feat(ai-markmap-agent): simplify writer prompt, fix resume and debug output

Writer prompt simplification:
- Remove URL format instructions from writer_behavior.md
- Compress problem data to ID list only (reduces tokens)
- Add _convert_plain_leetcode_to_links() to post_processing.py

Resume functionality fixes:
- Fix load_expert_responses_from_run to scan correct file categories
- Add parse_suggestions_from_response() and parse_adoption_list_from_response()
- Fix run_expert_review/run_full_discussion resume to load data into state

Debug output fixes:
- Add save_expert_review() for Phase 1 outputs
- Add save_discussion() for Phase 2 outputs
- Update graph.py to use new debug save methods

Author: lufftw

docs/pages/mindmaps/neetcode_ontology_agent_evolved_en.html

@@ -175,7 +175,7 @@ def save_optimizer_suggestion(
         round_num: int,
         output_key: str,
     ) -> Path | None:
-        """Save individual optimizer suggestion."""
+        """Save individual optimizer suggestion (legacy method)."""
         config = self.phases_config.get("optimization", {})
         if not config.get("save_optimizer_suggestions", False):
             return None
@@ -187,6 +187,55 @@ def save_optimizer_suggestion(
             f"round{round_num}",
         )
 
+    def save_expert_review(
+        self,
+        response: str,
+        expert_name: str,
+        suggestions: list | None = None,
+    ) -> Path | None:
+        """Save Phase 1 expert review output."""
+        config = self.phases_config.get("expert_review", {})
+        if not config.get("save_each_expert", False):
+            return None
+        
+        expert_key = expert_name.lower().replace(" ", "_")
+        
+        # Save raw response
+        self.save(1, "expert_review", response, expert_key)
+        
+        # Save suggestions as JSON if provided
+        if suggestions and config.get("save_all_suggestions", False):
+            suggestions_data = [s.to_dict() if hasattr(s, 'to_dict') else s for s in suggestions]
+            return self.save(1, "expert_suggestions", suggestions_data, expert_key, extension="json")
+        
+        return None
+    
+    def save_discussion(
+        self,
+        response: str,
+        expert_name: str,
+        adoption_list: list | None = None,
+    ) -> Path | None:
+        """Save Phase 2 discussion output."""
+        config = self.phases_config.get("discussion", {})
+        if not config.get("save_votes", False):
+            return None
+        
+        expert_key = expert_name.lower().replace(" ", "_")
+        
+        # Save raw response
+        self.save(2, "discussion", response, expert_key)
+        
+        # Save adoption list if provided
+        if adoption_list is not None and config.get("save_adoption_lists", False):
+            adoption_data = {
+                "expert": expert_name,
+                "adopted_ids": adoption_list if isinstance(adoption_list, list) else adoption_list.adopted_ids if hasattr(adoption_list, 'adopted_ids') else [],
+            }
+            return self.save(2, "adoption_list", adoption_data, expert_key, extension="json")
+        
+        return None
+    
     def save_summarizer_output(
         self,
         content: str,

docs/pages/mindmaps/neetcode_ontology_agent_evolved_zh-TW.html

@@ -415,11 +415,10 @@ def run_expert_review(state: WorkflowState) -> WorkflowState:
 
                 # Save debug output
                 if debug.enabled and expert.agent_id in state.get("expert_raw_responses", {}):
-                    debug.save_optimizer_suggestion(
+                    debug.save_expert_review(
                         state["expert_raw_responses"][expert.agent_id],
                         expert.name,
-                        1,
-                        "review"
+                        suggestions,
                     )
             except Exception as e:
                 error_msg = f"Error in {expert.name}: {e}"
@@ -509,11 +508,10 @@ def run_full_discussion(state: WorkflowState) -> WorkflowState:
 
                 # Save debug output
                 if debug.enabled and expert.agent_id in state.get("discussion_raw_responses", {}):
-                    debug.save_optimizer_suggestion(
+                    debug.save_discussion(
                         state["discussion_raw_responses"][expert.agent_id],
                         expert.name,
-                        2,
-                        "discussion"
+                        adoption_list,
                     )
             except Exception as e:
                 error_msg = f"Error in {expert.name} discussion: {e}"