fix: capture pending_user_turn before _restore_runtime_checkpoint clears it

_restore_runtime_checkpoint() clears both the runtime_checkpoint and
pending_user_turn flags (lines 1966-1967). recover_stale_sessions()
checked had_pending AFTER calling it, so had_pending was always False
when a checkpoint existed. This meant the re-publish of the last user
message never happened — the user got a notification but the interrupted
turn was never re-triggered.

nanobot/agent/loop.py

@@ -2063,7 +2063,12 @@ class AgentLoop:
             try:
                 session = self.sessions.get_or_create(key)
                 changed = False
-                had_pending = False
+                # Capture pending flag before _restore_runtime_checkpoint
+                # clears it (it clears both checkpoint and pending_turn at
+                # lines 1966-1967).
+                had_pending = bool(
+                    session.metadata.get(self._PENDING_USER_TURN_KEY)
+                )
                 if self._restore_runtime_checkpoint(session):
                     changed = True
                 if self._restore_pending_user_turn(session):