#     11 heavy/duplicate-family files demoted to tests/extended_suite.txt (they run
#     Forensic provenance: docs/testing/TEST_ARCHITECTURE.md + scripts/testing/verdicts.
#     breaker, financial batch salvage + dead-letter were heavy-only until now;
#     new behavioral batteries written by this mission (see the test files).
#     on every main push + release SHA — never dropped, only moved off the PR lane).
#     purge/embargo semantics) exercised on every push, not only heavy runs;
#   "RANGING") while every producer emits RegimeType values or nothing at all
#   'min_samples_to_reject' via 94ecf2c7->#199; 'forwarded', 'non-finite' at
#   (26,161 byte-identical rows, ~80 MB news.db, lane-04 RC2 + lane-06 §3).
#   (experience-paralysis, lane-08 §4). REJECT demotes to WARN below the shared
#   (live p50=$0.23, guardian kills median $0.30; 24% of decisions, RC-2).
#   -> ALL 14 setup families permanently NO_GO(REGIME_NOT_OK(UNKNOWN)) on
#   ACCEPT infinity, so TickData(bid=1,ask=inf) -> spread inf and (inf,inf) ->
#   BUG-223 NEXUS_AUDIT_DB seam -> tmp-path rows stamped CHAMPION). Pins:
#   BUG-278 (2026-09-14 wave): repo-root pytest clobbered the PRODUCTION
#   BUG-279 (2026-09-14 wave): feed-stall FAIL-LOUD — the 09-11 20:00Z freeze
#   BUG-280 (2026-09-14 wave): the early-return order-frequency throttle was a
#   BUG-281 (2026-09-14 wave): HIGH_SPREAD_CHOP chop band ($0.25/$0.18,
#   BUG-282 (2026-09-14 wave): RSS pubDate is RFC-822 but the parser accepted
#   BUG-283 (2026-09-14 wave): the hunter regime gate spoke ("TRENDING",
#   BUG-284 (2026-09-14 wave): Phase-09 suitability REJECT fired on 1-4 sample
#   BUG-285 (2026-09-14 wave, input-validation lane L11-1/2): pydantic gt/ge
#   BUG-286 FINDING-2 class guard (structural, role-11): no zero-byte
#   BUG-287 (role-11, 2026-09-15): PERF-WAVE R1 — the financial audit enqueue
#   DEGRADED via runtime_risk_state), weekend suppression, recovery reset,
#   EVALUATING floor (5, = StrategyEvaluator.min_samples_evaluating); at/above
#   FREEZE_ALL (fail-closed direction kept).
#   HARDCODED 60s floor pre-empting the configured cooldown_seconds=4.0 that
#   NaN passed the MUST-FAIL-CLOSED price boundary; the broker-history reader
#   Pins: RFC-822 parses to true event time, honest None + published_at_source
#   RED-BEFORE 3/3 at 65d3cb50. test_obs_trace_chain (the OBS-TRACE decision
#   Recalibrated to $0.45/$0.30 from the live broker window with machine
#   The 60-s publish bucket inside the article-hash identity therefore
#   UNREGISTERED at main — gate gap closed with the same wave it now tests.
#   adapter + regime stamped into retrain records; UNKNOWN/CHOP/FREEZE stay
#   agent17 OOS hard gate -> model-promotion OOS gate (empty-population refusal,
#   cache -> EXEC_TRACE rate limit; plus fail-closed / LIVE-PAPER guard pins.
#   calibrated on the 2026-05..08 p50=$0.04 era) froze NORMAL live spreads
#   champion bundle (force_fresh mint + load_database_config ignoring the
#   contract battery incl. the re-pinned unknown-regime diagnostic) was found
#   cooldown policy; pins strict-config still blocks, no magic literal.
#   duplicate-tick early-return skipped the whole pipeline. Pins: engine-owned
#   env-seam precedence, governed-tree force_fresh refusal, quarantine
#   evidence JSON pinned against the code defaults; extreme spreads still
#   families - the evidence contract was inverted (zero evidence passes, ONE
#   forwarded NaN/inf/inverted-OHLC bars into the canonical series (validation
#   gate-first skeleton heal -> dead-letter bounded retention -> /health TTL
#   identifier-shaped file may ever be TRACKED at repo root — three incidents
#   ineligible (fail-closed); adapter domain pinned against the enum.
#   live_freshness gate -> freshness characterization battery
#   loss blocks): 473/1373 decisions (34.5%) died at SUITABILITY_BELOW_THRESHOLD
#   memory; RED-BEFORE proven at 701bc50c.
#   migration row-scoping + idempotency, src call-site ban.
#   must never block the tick path longer than 100ms (old 2s floor), and R9 —
#   no raw stdout print on the decision path. Behavior pins + source pins;
#   non-finite BarData refused, get_historical_bars drops malformed rows loudly.
#   of cmd-hook '<'-sweep sentinels reaching commits ('the',
#   only ISO-8601 and silently fabricated now() on failure (88.1% of rows).
#   position_state_machine + recovery_budget -> qa_deep state battery
#   provenance, poll-stable timeless identity, legacy-DB column heal.
#   provider_gate -> qa_deep provider-gate chaos battery
#   ran 2.5 days: watchdog warned 1828x + re-emitted a HIGH incident every 15s
#   re-minted per poll -> dedup/tombstone guards structurally unable to fire
#   real data (lane-03 §3, verified 500 elite rows -> 0 GO). Pure vocabulary
#   runtime_safety battery -> P0 halt/kill-switch persistence, hot-path circuit
#   sizing fail-closed + protection ledger invariants + audit flush contract ->
#   stall episode (one START, one CRITICAL escalation past 900s grace ->
#   swallowed evaluations vs 1,469 decisions (96:1, RC-4). One authoritative
#   the #204 head). Predicate unit pins + live-tree gate + .gitignore guard
#   the floor the full NO_TRADE rejection semantics are pinned UNCHANGED.
#   the post-policy COOLDOWN gate reads (same _last_signal_time field): 78,798
#   update_engine (orchestrator/discovery/rollback) -> phase17 + BUG-237
#   was report-only). Pins: allow_inf_nan=False on all tick price/volume fields,
#   watchdog wiring reads (last-fresh stamp only on NEW ticks).
#   while runtime_risk_state() kept claiming RUNNING and the BUG-169
#  CR-007 registry entry records its protection. The comment above stays as the
#  historical provenance the role-7 wave wrote into this file.)
#  — measured 1252s aggregate; it now runs on main push + release SHA, and the
# (BarHandler/RuntimeLoop/TickPipeline/ProtectionEngine) targeting
# (monotonic stamps vs the caller's wall now_t) -> deep index scans every light
# (native MT5 history/tick windows + the live maintenance-window entry guard),
# (perf-wave R2) — stranded rows never returned to the ledger, dir grew unbounded.
# (reconstruction 2026-09-14: the battery itself moved to tests/extended_suite.txt
# (risk-engine ctor vs RiskConfig.max_allowed_lots).
# --- wave-2 (2026-09-15): CR-named owners + mutation-proven P0 contracts promoted;
# --- wave-2: CI-integrity meta-gates (fast, they police the tier system itself) ---
# ---------------------------------------------------------------------------
# 2026-09-16): a clean GitHub-release install died on launch with
# A2 dataset/feature/label lineage bounded probe (Agent-2 dataset forensics, 2026-09-05):
# AGENT-15 (2026-09-09): capital-protection boundary battery — exposure/spread
# AGENT-2 OBSERVABILITY GUARDRAILS (2026-09-01): executable enforcement of the frozen log contract — storm bounds, redaction, dormant-worker, recovery, evidence preservation (docs/architecture/observability-log-contract.md). Full list + rationale: tests/critical_suite_observability.txt
# AGENT4-SPRINT (2026-09-01): winner-protection geometry repair - BE trigger R-floor, lock offset 0.6p, tier floors 0.60/0.70/0.80 (regression suite)
# ArtifactIntegrityError LOAD_REJECTED (model.pt missing) -> PYI abort, because
# BREAKING/STALE encodings (4.0/5.0) exceeded the [-3,+3] family bound; producer-side
# BUG-165: BUG-140 terminal-outcome P0 regression net (Hermes-DevOps, 2026-08-31)
# BUG-166 (learning-loop battery): core experience/dataset/contract suites ran in NO pipeline
# BUG-183: purge/embargo leakage-guard defaults wired into the production research path (Hermes-DevOps, 2026-09-01)
# BUG-217 (Nexus-Main, 2026-09-02): 70D news state-encoding latent bound overflow -
# BUG-256 (Agent-15 capital protection, 2026-09-11): persisted-safety-state
# BUG-257 remainder (NSE-Swarm role 5, 2026-09-13): the P0-2 trust anchor
# BUG-258 (Hermes-Swarm, 2026-09-11): news DB list caps silently truncated
# BUG-264 (NSE-Swarm CI/GitOps, 2026-09-12): maintenance-window test-pin
# BUG-266 (NSE-Swarm role 1 Architect, 2026-09-13): the shipped paper REPLAY
# BUG-268 (NSE-Swarm role 2 Backend/Domain, 2026-09-13): minimum-loss grace
# BUG-269 (NSE-Swarm role 3, 2026-09-13): docker starter provisioner mints
# BUG-271 (NSE-Swarm role 4, 2026-09-13): governed in-place serving-artifact
# BUG-274 (NSE-Swarm role 6, 2026-09-14): bound-seam wrapper-state leak class.
# BUG-275 (NSE-Swarm role 6, 2026-09-14): hygiene cadence mixed clock domains
# BUG-276 (NSE-Swarm role 7, 2026-09-14): baseline-skeleton CONSTRAINT shadow -
# BUG-277 (NSE-Swarm role 9, 2026-09-14): doctor WARNING-actions filter matched
# BUG-285 (NSE-Swarm role 10, 2026-09-15) — ID renumbered: swarm reserved BUG-278 mid-PR, Hermes wave took it: durable financial-overflow was write-only
# BUG-288/289 (NSE-Swarm role 12, 2026-09-15) — audit-worker queue-adoption race (windows CI red
# BUG-290/291 (NSE-Swarm role 1, 2026-09-15): perf-wave R7+R8 unbounded in-process guard
# BUG-292 (NSE-Swarm role 2, 2026-09-15): perf-wave R6 — the C3 session spread-percentile
# BUG-293 (NSE-Swarm input-validation lane, 2026-09-15): lane-11 L11-4 + L11-6
# BUG-294 (NSE perf-bounds lane, 2026-09-15): behavioral regression net for the
# BUG-296 (NSE-Swarm zero-state lane, 2026-09-15): nexus repair --model is the
# BUG-297 (NSE-Swarm hot-path-sqlite lane, 2026-09-16): perf-wave R5 — the evaluator's
# BUG-298 (packaged fresh-install START crash + first-run model acquisition,
# BUG-299 (NSE-Swarm role4, 2026-09-16): the canonical-costs research bridge left
# BUG-303 (NSE-Swarm role 9 Observability, 2026-09-17): CI env probe, pip repair,
# BUG-305 (NSE-Swarm role 10 Accounting/Forensics, 2026-09-18): audit G1
# BUG-306 (NSE-Swarm role 11 CI/GitOps, 2026-09-18): Windows CI critical suite
# CHG-0048 (Nexus-Main UX, 2026-09-02): client UX safety contracts - mode-switch
# COVERAGE FORENSIC MISSION (2026-09-09): gate-closure, not bloat.
# Every push now gates the fetch->clean->train + dataset hardening contracts
# Every push proves the system still wires together (user brief: one file, full chain).
# First Setup & Local Training Environment Provisioning
# G3/BUG-259/BUG-260) - conservative zero + rate-limited HOLD_AGE_FALLBACK.
# I/O). Now an off-loop-refreshed in-process sketch with SQL-parity guarantees.
# ML-ARCH-001 (ScalpNet 3-Class Head Sunset — legacy-4 surface audit evidence)
# ML-CI-002 (ML Contract & Canonical Documentation Drift Prevention CI Gate)
# ML-DATA-001 (Historical M1 Market Data Ingest & Parquet Storage Harness)
# ML-EXP-001 (Immutable Experiment Registry & Artifact Manifest Schema)
# ML-EXP-003 (Bounded Hyperparameter Grid Search Runner)
# ML-FEAT-001 (50D Normalization End-to-End Parity Verification)
# ML-FEAT-002 (Feature Importance, Collinearity Clustering & Redundancy Pruning)
# ML-GOV-001 (Model Promotion Pipeline Pre-Flight & OOS Economic Gate Verification)
# ML-LABEL-002 (Sample Uniqueness Weighting & Label Overlap Anti-Leakage)
# ML-OBS-001 (Live Shadow Outcome Real-Time Resolution and Holding Metrics Wiring)
# ML-PHASE1-DATA-CONTRACT (2026-09-21): the 70D canonical data contract,
# ML-PLAT-001 (MT5 Runtime Boundaries & Remote Gateway Linux Parity Verification)
# ML-PLAT-002 (Signed Official Model Bundle Verification & Staging Slot Isolation)
# ML-RISK-001 (Position Replay & Economic Dataset Generation Engine)
# ML-TRAIN-001 (Deterministic Training Engine, Seed Harness & AMP Precision)
# ML-TRAIN-002 (Loss Function Exploration: Focal / Label Smoothing / Class-Balanced / Dice)
# ML-TRAIN-003 (Optimizer & LR-Schedule factory)
# ML-UI-001 (Deep Learning & Neural Model Studio Verification — 50D/70D)
# ML-UI-002 (Model Studio E2E Dataset Ingestion, 50D/70D Normalization & Position-Dataset Pipeline)
# ML-UI-003 (AI Hub: Model Registry, SQLite Catalog & Live Runtime Hot-Loader)
# ML-VAL-001 (Purged Walk-Forward Fold Monotonicity & Embargo Boundary Audit)
# ML-VAL-002 (Probability Calibration & Expected Calibration Error Evaluation)
# MT5-inspired Truthful Backtest Report Engine & UI contract
# MaintenanceCycle.run_cycle (2-sighting confirm, throttled, never blocks).
# Membership is evidence-driven: every file answers "what real defect does it
# Now one reused handle per (thread, timeout tier) through _connect_sqlite.
# OBS-002 (2026-09-09): correlation ids (EXEC-/INC-/upd-/fh-/req_) are the
# OBS-006 (2026-09-10): update lifecycle leaves log evidence — every stage
# ONLY supported zero-state model bootstrap (E1/E4/E8) — sidecar-stamped mint,
# P1/L5-L8/S10 extraction left 13 getattr(self, ...) reads in BOUND wrappers
# PAPER/LIVE isolation, duplicate execution, DB economic integrity, model
# PERF-DEADLETTER / PERF-HEALTH / PERF-EXEC-TRACE wave (2026-09-10):
# PERF-WAVE R4 (NSE-Swarm role 6, 2026-09-16): remote gateway recurring tick
# QA HARDENING MISSION (2026-09-07): critical-source floor mapping.
# RAW->CLEAN->FEATURES->LABELS->DATASET->MODEL INPUT chain on a 5k-bar M1 tail via the canonical builder (BUG-246 gap-constant SSoT pin).
# Registry of protected regressions: docs/testing/critical_regressions.md
# SELF-UPDATE CLI AUDIT (2026-09-10): real-CLI operator matrix found verify exit-code
# SMOKE-E2E (NexusMain, 2026-09-05): whole-application chain in ONE hermetic file
# TASK-STORAGE-HYGIENE (2026-09-09): 10GB client-disk mission — storage policy
# TIER 1 — PR/PUSH CRITICAL GATE (NSE radical test reconstruction 2026-09-14)
# These suites existed but ran in NO CI gate, so the contract they pin was
# Tick->50D->70D->ScalpNet->Policy->Risk->Paper dispatch->Audit DB->API v1
# USER ACTION / NEXT guidance surface was permanently dead. Filter fix +
# WAVE-2 ECOSYSTEM GATING (2026-09-05 forensic GAP-closure, QA verdict P0)
# Worker idle-pass drain + file cap + poison dead-letter + recovery observability.
# artifact contract) are mutation-validated: scripts/testing/mutation_check.py
# b1fe0137; pins live in the already-gated test_outcome_flush_race_bug140.py) + telegram_notify CLI
# behavior, not just its structure).
# blocking fresh-install START smoke in release.yml.
# boot-#2 load, CHAMPION/verified-bundle clobber refusals (E9), db-migrate lock
# bounded-runtime class — eviction driven through the REAL dispatch seam, broken-
# call-shape break (OS-leg + JS-lane finish notifications never delivered).
# candle-intel/MSIE bar feeds + live chart overlays + breakeven-breach check
# catch?" and runs deterministic + fast. Measured cost outliers moved to
# collision — pins a provably-outside-window timestamp so the dispatch
# composition-root attributes -> permanent defaults: PERF-02 5s throttles never
# confirmation gate (BUG-194), connectivity truth, palette safety, i18n/RTL,
# contract breaks (VERIFIED -> 5, tamper -> 5) + transient-cache FAIL + --json stdout
# correlation id (incl. the -<hex8> uuid suffix the redactor previously ate).
# cycle (12x intended IO). Wall-domain contract + RED-before pins.
# decision humanization. 28 tests over real assets + create_app serve surface.
# derived-registry reads (pre-trade tier-3 fallback twin of BUG-292) paid a raw
# dies permanently after a legitimate retrain. Champion supersession seam.
# engage (per-tick gateway RPC), hourly online-train gates fire per bar,
# engine (allowlist cleanup), runtime retention re-prune, MT5 failure-storm
# engine runs (the 09-11 bb1f0afe shape) serves poisoned weights silently
# every default-costs backtest paid 25% of real cost and the robustness GATE8 stress
# evidence; a cap-pinned gate fails closed ROBUSTNESS_NOT_SIMULABLE.
# explicit caller limits (newest-500) -> dataset news-frame export lost the
# fail-open boundaries — paper_state.json partial-apply on corrupt reload and
# fingerprint -> P0-2 boot trust anchor refuses the NEXT cold boot; trading
# fires at BOOT only; a foreign process re-landing drifted bytes while the
# gate ever launched the built EXE in a start shape. Adds the model_provisioning
# gate must never fall back to the host wall clock (clock-domain parity with
# gate ran a per-call SQLite connect + same-day SELECT INSIDE policy evaluation (loop-thread
# gate-first fresh DBs dead-letter every ON CONFLICT producer forever (15 of 17
# gateway order payloads silently defaulting volume/price before broker dispatch.
# heartbeat, lane report, and telegram notify entry point robustness.
# idempotency / safe-mode batteries can never re-flake at a wall-clock minute.
# labeling identity, fail-closed serving validator, provenance census, and
# log<->DB join key and must survive the high-entropy redactor; secrets stay
# maps (_processed_orders, _ai_flip_warn_times, AccountingCore._report_cache) -> BoundedLRUMap.
# max_slippage_ticks at the 5.0 model default while calibrated friction is ~20 ticks —
# mkdir (E6) and the cp1252 pre-bind panel safety (W12).
# mode had zero production wiring (audit K2) - pins the paper-data factory as
# module to have at least one critical-suite guardian. These entries map
# not just the epoch-conversion helper. Class-wide AST guard included.
# nothing provisioned a serving bundle for an INSTALLED release and no release
# oldest analyzed history; windowed exports dropped rows INSIDE the window.
# on cancel, paper/direct inline, execution untouched).
# on the real OrderLifecycleManager composition root.
# package (PATH A signed official download / PATH B local training), packaged
# parity — official_install._sync_file os.fsync on "rb" fails with EBADF (CRT _commit
# persist (async retrain / collapse recovery / hot-swap) orphans the CHAMPION
# pollution; pinned by the contract battery below.
# probe -> fresh-volume container restart loop. Servability-gated re-mint.
# read offloaded from the event loop in RuntimeLoop._poll_tick (shielded, drained
# redacted on the same path (artifacts/forensics/observability-audit.json).
# remainder — the runtime broker-offset override must reach EVERY consumer
# repair mirrors the dataset clamp semantics. Companion to BUG-190/BUG-197.
# requires write access); test_model_studio.py charmap UnicodeDecodeError on UTF-8 Web/index.html.
# scenarios were exact no-ops (PASS with max_degradation=0.0). Cap now derived from
# scripts/ci/critical_manifest.py REQUIRES every execution-critical
# shadow70/news-coverage gates landed by agent/feature/ML-PHASE1-DATA-CONTRACT.
# sqlite3.connect per call on the loop thread and bypassed the repository's URI contract.
# squeeze limits, dispatch idempotency, clamp insanity guards, config-contract pin
# stdout log_signal pin, report-cache LRU under real browsing (pins the BUG-287/290/291
# targets). Unique-index heal at baseline + boot + class-wide ON-CONFLICT guard.
# tests/extended_suite.txt (Tier 2, main-push). P0 safety domains (risk math,
# that were landed but not executed in any CI gate until this wave.
# the previously-unguarded floor modules (drift closure, not bloat):
# the rendered label "WARN" against the producer vocabulary "WARNING" -> the
# the single PAPER construction path + fail-closed / loud-degrade semantics.
# throttle, protected-data + concurrency + growth-clamp regression nets.
# transition + blocked/failed/rollback verdict logs [UPDATE] with the upd-
# unreachable. RED-BEFORE 10/10 proven at fd95a7b0.
# until the next refusal. Alert-only champion-drift sentinel wired into
# untrained weights that pass the digest gate then die on the P0 serving
# unverified on every push.
# vocabulary contract + class-wide verdict=="WARN" comparison guard.
# wiring on the dispatch layer — the HALTED/KILL_SWITCH gate must stay live
# workspace+configs anchoring, the labeled DEV STARTER fallback, and the
tests/ci/test_beforepush_wrapper.py
tests/ci/test_check_workflows.py
tests/ci/test_classify_changes.py
tests/ci/test_critical_manifest_gate.py
tests/ci/test_gate_scope_integrity.py
tests/ci/test_mutation_runner_smoke.py
tests/cli/test_docs_consistency.py
tests/e2e/test_smoke_chain.py
tests/integration/test_database_execution_audit.py
tests/integration/test_model_promotion_pipeline_e2e.py
tests/integration/test_mt5_accounting_api_contract.py
tests/integration/test_mt5_adapter_parity.py
tests/integration/test_official_model_tamper_proofing.py
tests/integration/test_signal_pipeline_health.py
tests/release/test_release_hardening.py
tests/unit/news/ingest/test_bug282_rfc822_and_identity.py
tests/unit/test_50d_normalization_parity.py
tests/unit/test_70d_bug106_incremental_phase19.py
tests/unit/test_70d_contract_parity_task3.py
tests/unit/test_70d_dataset_parity_task3.py
tests/unit/test_70d_inference_validator_task3.py
tests/unit/test_70d_parity_task3.py
tests/unit/test_70d_replay_parity_task3.py
tests/unit/test_a2_data_lineage_bounded.py
tests/unit/test_accounting_advanced_metrics.py
tests/unit/test_accounting_deduplication.py
tests/unit/test_accounting_pnl_regression.py
tests/unit/test_accounting_timezone.py
tests/unit/test_agent11_execution_risk_forensic.py
tests/unit/test_agent12_execution_forensic.py
tests/unit/test_agent14_dataset_integrity.py
tests/unit/test_agent16_walkforward_purge_embargo_leakage.py
tests/unit/test_agent17_oos_hard_gate.py
tests/unit/test_agent5_bug249_251_252.py
tests/unit/test_agent6_execution_contract.py
tests/unit/test_audit_flush_contract.py
tests/unit/test_audit_legacy4_surface.py
tests/unit/test_backtest_report.py
tests/unit/test_backtest_report_asset.py
tests/unit/test_backtest_report_integration.py
tests/unit/test_bug185_terminal_state_resolver.py
tests/unit/test_bug217_news_state_bounds.py
tests/unit/test_bug225_untrained_champion_canary.py
tests/unit/test_bug232_mode_boundary.py
tests/unit/test_bug237_update_setup_payload.py
tests/unit/test_bug254_executions_idempotency_migration.py
tests/unit/test_bug256_dispatch_halt_wiring.py
tests/unit/test_bug257_champion_drift_sentinel.py
tests/unit/test_bug258_reversal_risk_bypass.py
tests/unit/test_bug259_breaker_anchor_restart.py
tests/unit/test_bug260_margin_usage_enforcement.py
tests/unit/test_bug261_terminal_outcome_tick_domain.py
tests/unit/test_bug262_close_time_evidence.py
tests/unit/test_bug264_maintenance_pin_collision.py
tests/unit/test_bug266_paper_data_wiring.py
tests/unit/test_bug268_min_loss_grace_clock.py
tests/unit/test_bug269_docker_starter_servable.py
tests/unit/test_bug271_champion_retrain_fingerprint.py
tests/unit/test_bug274_wrapper_state_leak.py
tests/unit/test_bug275_hygiene_cadence_clock.py
tests/unit/test_bug277_doctor_warning_vocabulary.py
tests/unit/test_bug278_shadow70_not_compared.py
tests/unit/test_bug278_test_harness_artifact_isolation.py
tests/unit/test_bug279_feed_stall_fail_loud.py
tests/unit/test_bug280_frequency_throttle_unification.py
tests/unit/test_bug281_chop_boundary_recalibration.py
tests/unit/test_bug283_hunter_regime_adapter.py
tests/unit/test_bug284_suitability_micro_sample_floor.py
tests/unit/test_bug285_overflow_drain.py
tests/unit/test_bug285_overflow_wiring.py
tests/unit/test_bug285_price_boundaries_fail_closed.py
tests/unit/test_bug286_sentinel_class_guard.py
tests/unit/test_bug289_telegram_notify_cli.py
tests/unit/test_bug290_bounded_guard_maps.py
tests/unit/test_bug292_spread_sketch_offload.py
tests/unit/test_bug293_failopen_boundaries.py
tests/unit/test_bug294_bounded_runtime_behavior.py
tests/unit/test_bug296_repair_model.py
tests/unit/test_bug297_evaluator_hot_path_reads.py
tests/unit/test_bug298_fresh_install_bootstrap.py
tests/unit/test_bug298_model_provisioning.py
tests/unit/test_bug298_provisioning_web_api.py
tests/unit/test_bug299_friction_cap_saturation.py
tests/unit/test_bug300_ai_triage_fallback.py
tests/unit/test_bug300_workflow_ai_wiring.py
tests/unit/test_bug303_ci_notify_env_and_report.py
tests/unit/test_bug305_broker_offset_env_consumer_parity.py
tests/unit/test_bug306_windows_ci_compat.py
tests/unit/test_byte_identical_news_dedup.py
tests/unit/test_calibration_research.py
tests/unit/test_capital_protection_a15.py
tests/unit/test_champion_writer_governance.py
tests/unit/test_ci_final_gate_fail_closed.py
tests/unit/test_confidence_calibration.py
tests/unit/test_critical_suite.py
tests/unit/test_data_ingest.py
tests/unit/test_dataset_manifest.py
tests/unit/test_dataset_split_purge_bug244.py
tests/unit/test_disk_safety_guards.py
tests/unit/test_domain_models.py
tests/unit/test_economic_parity_harness.py
tests/unit/test_evidence_semantics_bug140.py
tests/unit/test_execution_architecture.py
tests/unit/test_experiment_registry.py
tests/unit/test_factory_backtest_report_projection.py
tests/unit/test_feature_importance.py
tests/unit/test_gap_safe_sequences.py
tests/unit/test_gate_convergence_evidence.py
tests/unit/test_hardened_protocol.py
tests/unit/test_hardening_p1_p2.py
tests/unit/test_hot_swap_governance.py
tests/unit/test_hunter_metadata_persistence.py
tests/unit/test_hunter_min_quality_pins_bug227.py
tests/unit/test_hunter_sample_strategy_coverage.py
tests/unit/test_hyperparam_search.py
tests/unit/test_label_integrity.py
tests/unit/test_lifecycle_bug140.py
tests/unit/test_lifecycle_event_projection.py
tests/unit/test_lifecycle_verdict_mapping.py
tests/unit/test_liquidity_engine_causality.py
tests/unit/test_liquidity_engine_contract.py
tests/unit/test_live_freshness_characterization.py
tests/unit/test_live_shadow_outcome_resolution.py
tests/unit/test_logging_correlation_ids.py
tests/unit/test_logging_redaction.py
tests/unit/test_loss_functions.py
tests/unit/test_maintenance_dispatch_guard.py
tests/unit/test_market_data_integrity_ag13.py
tests/unit/test_marketplace_backend.py
tests/unit/test_ml_contract_drift.py
tests/unit/test_ml_phase1_census_consistency.py
tests/unit/test_ml_phase1_data_contract.py
tests/unit/test_ml_phase1_labeling_identity.py
tests/unit/test_ml_phase1_serving_contract.py
tests/unit/test_model_registry_hot_load.py
tests/unit/test_model_studio.py
tests/unit/test_model_studio_dataset_order.py
tests/unit/test_model_studio_pipeline.py
tests/unit/test_mslie_sweep_window_pins_bug227.py
tests/unit/test_mt5_diag_throttle.py
tests/unit/test_bug307c_cli_logging_disable_leak.py
tests/unit/test_news_bridge_contract_phase13b.py
tests/unit/test_news_bridge_finalize_phase13b.py
tests/unit/test_nse_lifecycle_regression_matrix.py
tests/unit/test_obs_trace_chain.py
tests/unit/test_optimizers_schedulers.py
tests/unit/test_order_lifecycle.py
tests/unit/test_order_manager_exit_bugs.py
tests/unit/test_order_write_uncertainty.py
tests/unit/test_outcome_flush_race_bug140.py
tests/unit/test_p0_producer_regression.py
tests/unit/test_paper_live_training_lineage.py
tests/unit/test_paper_persistence.py
tests/unit/test_paper_replay_mode.py
tests/unit/test_pending_recovery_cause_aware.py
tests/unit/test_perf_health_probe_cache.py
tests/unit/test_perf_r1_blocking_put_bound.py
tests/unit/test_perf_r4_runtime_loop_offload.py
tests/unit/test_perf_safety_contract_pins.py
tests/unit/test_performance_metric_truth.py
tests/unit/test_persistence_boot_trust.py
tests/unit/test_persistence_durability_guarantees.py
tests/unit/test_policy.py
tests/unit/test_policy_flip_protection_pins_bug227.py
tests/unit/test_policy_throttle_pins_bug227.py
tests/unit/test_policy_warmup_range_seam_bug230.py
tests/unit/test_position_replay_pipeline.py
tests/unit/test_predictive_limit_explainability_bug226.py
tests/unit/test_promotion_economic_integrity.py
tests/unit/test_promotion_rejects_degenerate_model.py
tests/unit/test_protection_ledger_invariants.py
tests/unit/test_purge_embargo_monotonicity.py
tests/unit/test_qa_deep_execution_safety.py
tests/unit/test_qa_deep_provider_gate_chaos.py
tests/unit/test_qa_deep_state_machines.py
tests/unit/test_quality_tier_pins_bug227.py
tests/unit/test_recon_batch_atomicity_real_failure.py
tests/unit/test_recon_dead_letter_store_contracts.py
tests/unit/test_recon_duplicate_deal_across_restart.py
tests/unit/test_recon_requote_and_reconcile.py
tests/unit/test_recon_risk_boundary_battery.py
tests/unit/test_recon_wal_crash_durability.py
tests/unit/test_regime_calibration_bug132.py
tests/unit/test_release_system.py
tests/unit/test_release_update_phase17.py
tests/unit/test_replay_toggle_guard.py
tests/unit/test_research_archive_round3.py
tests/unit/test_research_edge_hardening_20260909.py
tests/unit/test_research_history_surface.py
tests/unit/test_research_loop_smoke.py
tests/unit/test_research_purge_defaults_bug183.py
tests/unit/test_research_regime_report.py
tests/unit/test_research_selection_bias_20260909.py
tests/unit/test_research_task4_dataset.py
tests/unit/test_research_walkforward_gate_agent5.py
tests/unit/test_research_worker_multiplicity.py
tests/unit/test_risk_circuit_breakers.py
tests/unit/test_risk_engine.py
tests/unit/test_robustness_stress_scenarios.py
tests/unit/test_runtime_bars_normalize.py
tests/unit/test_runtime_config_hot_reload.py
tests/unit/test_runtime_safety_mission.py
tests/unit/test_sample_weights.py
tests/unit/test_schema_70d_reconciliation.py
tests/unit/test_shadow70_health_drift.py
tests/unit/test_shadow70_safety.py
tests/unit/test_shadow_hardening_chg0046.py
tests/unit/test_signed_update_manifest.py
tests/unit/test_sizing_policy_failclosed.py
tests/unit/test_smoke_self.py
tests/unit/test_storage_lifecycle_wiring.py
tests/unit/test_storage_policy.py
tests/unit/test_strategy_factory_phase22.py
tests/unit/test_strategy_factory_registry_invariant.py
tests/unit/test_survival_drawdown_runtime_bug132.py
tests/unit/test_temporal_liquidity_phase20.py
tests/unit/test_training_dataset_source.py
tests/unit/test_training_determinism.py
tests/unit/test_training_env_exact.py
tests/unit/test_training_env_web.py
tests/unit/test_training_env_worker.py
tests/unit/test_training_stage_progress.py
tests/unit/test_triple_barrier_parameters.py
tests/unit/test_update_cli_contract.py
tests/unit/test_update_lifecycle_logging.py
tests/unit/test_user_hunt_bug170_171.py
tests/unit/test_web_ux_safety.py
tests/unit/test_wft_fold_economics.py
tests/unit/test_wft_time_based_purge.py
tests/unit/test_winner_protection_geometry_sprint.py
tests/unit/test_zero_friction_guard_e1.py