refactor: refactor core framework to better propogate feedbacks (#599)

* refactor: Update type annotations and remove unused class in evolving modules

* refactor: Simplify evolving agent and feedback handling in CoSTEER module

* lint & CI

* mypy

* ruff for core

* mypy

* refactor: remove unnecessary comments and update feedback handling logic

* refactor: Add prev_task_feedback parameter to evolving strategies

* feat: Clear folder before extracting zip file in DockerEnv

* fix: Correct retrieval of last experiment from history

rdagent/core/developer.py

@@ -14,13 +14,21 @@ class Developer(ABC, Generic[ASpecificExp]):
         self.scen: Scenario = scen
 
     @abstractmethod
-    def develop(self, exp: ASpecificExp) -> ASpecificExp:
+    def develop(self, exp: ASpecificExp) -> ASpecificExp:  # TODO: remove return value
         """
         Task Generator should take in an experiment.
 
         Because the schedule of different tasks is crucial for the final performance
         due to it affects the learning process.
 
+        Current constraints:
+        - The developer should **inplace** edit the exp instead of returning value;
+            - because we have a lot of use cases to raise errors, but we need the intermediate results in exp.
+        - So we should remove the return value in the future.
+
+        Responsibilities:
+        - Generate a new experiment after developing on it.
+        - If it tries to deliver message for future development, it should set a ExperimentFeedback
         """
         error_message = "generate method is not implemented."
         raise NotImplementedError(error_message)