docs: annotate comparative review with implementation responses

Add [RESPONSE] annotations to each finding in the comparative review:
- Findings A, B, C, E: acknowledged as valid with action items
- Finding D (binary apply): noted as low priority / theoretical
- Finding F (simulation artifacts): corrected — reviewer checked raw
  LLM outputs (sim*-<name>/generated.py) instead of extracted files
  (sim*/todo.py etc.) which all pass py_compile

Also fix CommittingSessionService._commit() to log exceptions instead
of silently swallowing them (valid Finding E).

src/cleveragents/application/services/session_service.py

@@ -328,7 +328,11 @@ class CommittingSessionService(PersistentSessionService):
             sa_session = self._sa_factory()
             sa_session.commit()
         except Exception:
-            pass
+            import structlog
+
+            structlog.get_logger(__name__).warning(
+                "session commit failed", exc_info=True
+            )
 
     def create(self, actor_name: str | None = None) -> Session:
         result = super().create(actor_name)