perf: add object pooling and enhanced metrics for streaming optimization

- Implement gRPC/shred connection pools for memory efficiency
- Add atomic processing time stats with auto-calibration clock
- Optimize event parsers and protocol handlers
- Refactor metrics system for advanced performance monitoring

src/streaming/shred_stream.rs

@@ -8,7 +8,8 @@ use crate::protos::shredstream::SubscribeEntriesRequest;
 use crate::streaming::common::{EventProcessor, SubscriptionHandle};
 use crate::streaming::event_parser::common::filter::EventTypeFilter;
 use crate::streaming::event_parser::{Protocol, UnifiedEvent};
-use crate::streaming::shred::TransactionWithSlot;
+use crate::streaming::event_parser::core::traits::get_high_perf_clock;
+use crate::streaming::shred::pool::factory;
 use log::error;
 use solana_entry::entry::Entry;
 
@@ -57,10 +58,10 @@ impl ShredStreamGrpc {
                         if let Ok(entries) = bincode::deserialize::<Vec<Entry>>(&msg.entries) {
                             for entry in entries {
                                 for transaction in entry.transactions {
-                                    let transaction_with_slot = TransactionWithSlot::new(
+                                    let transaction_with_slot = factory::create_transaction_with_slot_pooled(
                                         transaction.clone(),
                                         msg.slot,
-                                        chrono::Utc::now().timestamp_micros(),
+                                        get_high_perf_clock(),
                                     );
                                     // 直接处理，背压控制在 EventProcessor 内部处理
                                     if let Err(e) = event_processor_clone