fix: reduce false container-failed notifications, add error details to UI

Backend (health_monitor.py):
- Skip health checks for instances with no container_id
- Treat 'not_found' as error only when container was previously running
- Skip duplicate error notifications when already in error state
- Skip 'not_found' notifications for containers that never ran

Frontend (notification-item.tsx):
- Display notification.message (detailed error text)
- Add expandable Details section showing metadata (exit_code, previous_status, etc.)
- New CSS styles for message and metadata display

Quality gates: py_compile, tsc --noEmit, 80/80 tests pass

apps/api/src/services/instance/health_monitor.py

@@ -90,8 +90,16 @@ class HealthMonitor:
         instance: ToolInstance,
     ) -> None:
         """Check a single instance and handle state transitions."""
+        # Skip instances that have never been assigned a container.
+        if not instance.container_id:
+            logger.debug(
+                "Skipping health check for instance %s: no container_id",
+                instance.id,
+            )
+            return
+
         try:
-            container_info = get_container_status(instance.container_id or "")
+            container_info = get_container_status(instance.container_id)
         except Exception:
             logger.exception(
                 "Health check failed for instance %s",
@@ -135,7 +143,11 @@ class HealthMonitor:
         previous = self._last_known_state.get(instance.id)
 
         # Determine new status
-        new_status = self._derive_status(snapshot)
+        new_status = self._derive_status(
+            snapshot,
+            previous,
+            instance.status,
+        )
 
         # If first check or state changed
         if previous is None or not self._snapshots_equal(previous, snapshot):
@@ -144,10 +156,34 @@ class HealthMonitor:
             )
             self._last_known_state[instance.id] = snapshot
 
-    def _derive_status(self, snapshot: HealthSnapshot) -> str:
-        """Derive instance status from health snapshot."""
+    def _derive_status(
+        self,
+        snapshot: HealthSnapshot,
+        previous: HealthSnapshot | None,
+        current_status: str | None,
+    ) -> str:
+        """Derive instance status from health snapshot.
+
+        Treats missing containers as an error only when the container was
+        previously known to be running. This avoids false "container failed"
+        alerts for instances that are still starting or have no container yet.
+        """
+        if snapshot.container_status == "not_found":
+            # If the container was never seen running, assume it's still
+            # starting or was deleted intentionally; don't flag as error.
+            if previous is None and current_status == "starting":
+                return "starting"
+            if previous is not None and previous.container_status == "running":
+                return "error"
+            # Fall back to current status to avoid spurious errors.
+            return current_status or "error"
+
+        if snapshot.container_status == "exited":
+            return "error"
+
         if snapshot.container_status != "running":
             return "error"
+
         if snapshot.tunnel_healthy is False:
             return "unhealthy"
         return "running"
@@ -223,6 +259,16 @@ class HealthMonitor:
         # Create notification for instance owner (fire-and-forget)
         # Only send warnings and errors; skip "recovered" info notifications.
         if new_status == "error":
+            # Skip duplicate error notifications if already in error state.
+            if previous_status == "error":
+                return
+            # Skip "not_found" errors for containers that were never running
+            # (e.g. still starting, or intentionally stopped/deleted).
+            if (
+                snapshot.container_status == "not_found"
+                and (previous is None or previous.container_status != "running")
+            ):
+                return
             category = "instance"
             severity = "error"
             title = "Container failed"