snapshot: WIP 5j replay reliability (B1 watchdog + dialog handlers + grounding drift)

Snapshot avant correction du blocage relance Léa (3 incidents 24h: SSH refusé,
polls morts ×2). Point de rollback stable.

Contenu:
- agent_v1/core/executor.py: 5 patchs dialog handling (saveas drift, close_tab
  hotkey fallback, confirm_save Unicode apostrophe, foreground dialog
  recontextualization, runtime_dialog in-loop) + helpers normalize_window_hint,
  requires_post_verify_window_transition
- agent_v1/core/grounding.py: garde drift template fix (fallback_x/y plumbed)
- server_v1/replay_watchdog.py (NEW): orphan watchdog B1, scan 10s timeout 30s
- server_v1/api_stream.py: dispatched_action plumbing, watchdog lifespan,
  metrics endpoint
- server_v1/replay_engine.py: _schedule_retry préserve original_action +
  dispatched_action
- stream_processor.py: gardes _infer_tab_switch_target (no false switch_tab
  on save_as dialog open) + _attach_expected_window_before
- tests/integration: test_replay_watchdog.py (8 cas), test_stream_processor.py
- tests/unit: test_executor_verify_window_guard.py (start_button, close_tab,
  runtime_dialog, post_verify, transition fallbacks)

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

agent_v0/agent_v1/main.py

@@ -28,6 +28,7 @@ from .ui.chat_window import ChatWindow
 from .ui.capture_server import CaptureServer
 from .session.storage import SessionStorage
 from .vision.capturer import VisionCapturer
+from .finalize_contract import dispatch_finalize_result
 
 # Import optionnel du client serveur (pour le chat et les workflows)
 # Deux chemins : relatif (depuis agent_v0.agent_v1) ou absolu (depuis C:\rpa_vision\agent_v1)
@@ -80,6 +81,7 @@ class AgentV1:
         self._executor = None
         # Flag pour indiquer qu'un replay est en cours (eviter les conflits)
         self._replay_active = False
+        self._last_recording_name = ""
 
         # Etat partage entre systray et chat (source de verite unique)
         self._state = AgentState()
@@ -210,12 +212,14 @@ class AgentV1:
             time.sleep(30)  # Vérifier toutes les 30s
 
     def start_session(self, workflow_name):
+        self._last_recording_name = workflow_name
         self.session_id = f"sess_{time.strftime('%Y%m%dT%H%M%S')}_{uuid.uuid4().hex[:6]}"
         self.session_dir = self.storage.get_session_dir(self.session_id)
 
         self.vision = VisionCapturer(str(self.session_dir))
 
         self.streamer = TraceStreamer(self.session_id, machine_id=self.machine_id)
+        self.streamer.set_on_finalize_result(self._on_finalize_result)
         self.captor = EventCaptorV1(self._on_event_bridge)
 
         # Initialiser l'executeur partage
@@ -325,6 +329,15 @@ class AgentV1:
                     # pour enchainer les actions du workflow
                     time.sleep(0.2)
                 else:
+                    if getattr(self._executor, "_replay_paused", False):
+                        if not self._replay_active:
+                            self._replay_active = True
+                            self.ui.set_replay_active(True)
+                            self._state.set_replay_active(True)
+                        poll_delay = getattr(self._executor, '_poll_backoff', REPLAY_POLL_INTERVAL)
+                        time.sleep(max(poll_delay, REPLAY_POLL_INTERVAL))
+                        continue
+
                     # Pas d'action en attente — utiliser le backoff de l'executor
                     # (augmente si le serveur est indisponible, reset a 1s sinon)
                     if self._replay_active:
@@ -429,6 +442,11 @@ class AgentV1:
             f"agent_{self.user_id}"
         )
 
+    def _on_finalize_result(self, payload: dict) -> None:
+        """Réagir au contrat enrichi de /finalize côté agent."""
+        replay_name = self._last_recording_name or "la tâche que vous venez d'enregistrer"
+        dispatch_finalize_result(self.ui, payload, replay_name)
+
     _last_heartbeat_hash: str = ""
 
     def _heartbeat_loop(self):