fix(agentex): recheck task status on keepalive to catch lost terminal events

Pure event-driven termination can't recover if a terminal transition's
task_updated publish fails (the DB flips terminal but the event never lands,
and the errors are logged-and-suppressed). On each keepalive interval, re-read
the task status and end the stream if it is terminal — an authoritative
backstop for a dropped event. Status-only, so it never closes a live task on a
reclaimed key.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: deepthi-rao-scale

agentex/src/domain/use_cases/streams_use_case.py

@@ -161,10 +161,17 @@ async def stream_task_events(
                     # healthy again, so reset the backoff/error counter.
                     consecutive_errors = 0
 
-                    # Idle: send keepalive ping so proxies don't reap us.
+                    # Idle: recheck status (catches a lost terminal event), else ping.
                     if message_count == 0:
                         current_time = asyncio.get_running_loop().time()
                         if current_time - last_message_time >= ping_interval:
+                            task = await self.task_service.get_task(id=task_id)
+                            if task.status in TERMINAL_TASK_STATUSES:
+                                logger.info(
+                                    f"Ending SSE stream for task {task_id}: "
+                                    "terminal on status recheck"
+                                )
+                                return
                             yield ":ping\n\n"
                             last_message_time = current_time
                         await asyncio.sleep(0.1)