WIP: memory system improvements (in progress)

Interview/chat prompt layers, reply planner, style profiles, memory
injection, interview meta store, and related tests. Work not finished.

Made-with: Cursor

api/app/agents/chat/prompt_context.py

@@ -31,14 +31,15 @@ class ChatPromptContext:
     turn_plan: InterviewTurnPlan | None = None
 
     def guided_system_prompt(self) -> str:
-        """用户原话仅以对话历史 + HumanMessage 注入模型。"""
-        from app.agents.chat.interview_turn_plan import (
-            format_interview_turn_directive_block,
-        )
+        """用户原话仅以对话历史 + HumanMessage 注入模型。
+
+        本轮模式硬指令统一由 `InterviewTurnPlan.render_system_directive()` 产出，
+        主 prompt 不得重复立法（见 interview_turn_plan 模块 docstring）。
+        """
         from app.agents.chat.prompts_conversation import get_guided_conversation_prompt
 
         directive = (
-            format_interview_turn_directive_block(self.turn_plan)
+            self.turn_plan.render_system_directive()
             if self.turn_plan is not None
             else ""
         )