fix: clean MeshCore adapter shutdown

close() stopped the asyncio loop without disconnecting first, leaving the
EventDispatcher._process_events() task and the serial transport pending
("Task was destroyed but it is pending!" / "Event loop is closed").
Now await MeshCore.disconnect() and let the transport's connection_lost
callback settle before stopping the loop, then join the thread.

Found by running the adapter against a live MeshCore-flashed Heltec V3.

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

host-bridge/adapters/meshcore_adapter.py

@@ -79,5 +79,25 @@ class MeshCoreAdapter:
         )
 
     def close(self) -> None:
-        if self._loop is not None:
-            self._loop.call_soon_threadsafe(self._loop.stop)
+        if self._loop is None:
+            return
+        # Disconnect on the loop thread first: MeshCore.disconnect() stops the
+        # EventDispatcher task and closes the serial transport. Skipping this
+        # and stopping the loop directly leaves _process_events() pending
+        # ("Task was destroyed but it is pending!" / "Event loop is closed").
+        if self._mc is not None:
+            async def _shutdown() -> None:
+                await self._mc.disconnect()
+                # Let the serial transport's connection_lost callback run to
+                # completion before the loop stops, otherwise it lingers as a
+                # pending SerialTransport._call_connection_lost task.
+                await asyncio.sleep(0.25)
+
+            try:
+                fut = asyncio.run_coroutine_threadsafe(_shutdown(), self._loop)
+                fut.result(timeout=5)
+            except Exception:
+                pass
+        self._loop.call_soon_threadsafe(self._loop.stop)
+        if self._thread is not None:
+            self._thread.join(timeout=5)