fix: harden L3 runtime continuity and tool execution

Align the L3 graph, agent service, and sync tool shims on one canonical continuity contract so clarification resumes and persisted snapshots behave consistently. Add targeted regressions and hardening notes covering system-message coalescing, async bridge usage, and continuity rehydration.

backend/app/agents/tools/time_reasoning.py

@@ -241,6 +241,10 @@ def normalize_tool_time_arguments(tool_name: str, args: dict, current_datetime_c
         if raw_value and not _is_iso_datetime(raw_value):
             payload = resolve_time_expression_data(raw_value, current_datetime_context=current_datetime_context, prefer="datetime")
             normalized["reminder_at"] = payload["resolved_datetime"]
+        raw_date = normalized.get("date")
+        if isinstance(raw_date, str) and raw_date.strip() and not _is_iso_date(raw_date):
+            payload = resolve_time_expression_data(raw_date, current_datetime_context=current_datetime_context, prefer="date")
+            normalized["date"] = payload["resolved_date"]
         return normalized
 
     if tool_name in {"create_schedule_task", "create_task"}: