fix: handle session closure during concurrent TMDB requests

- Re-ensure session before each request attempt to handle race conditions
- Add AttributeError handling for None session
- Detect 'Connector is closed' errors and recreate session
- Fixes AttributeError: 'NoneType' object has no attribute 'get' during concurrent NFO processing

src/core/services/tmdb_client.py

@@ -115,6 +115,12 @@ class TMDBClient:
         
         for attempt in range(max_retries):
             try:
+                # Re-ensure session before each attempt in case it was closed
+                await self._ensure_session()
+                
+                if self.session is None:
+                    raise TMDBAPIError("Session is not available")
+                
                 logger.debug(f"TMDB API request: {endpoint} (attempt {attempt + 1})")
                 async with self.session.get(url, params=params, timeout=aiohttp.ClientTimeout(total=30)) as resp:
                     if resp.status == 401:
@@ -133,8 +139,14 @@ class TMDBClient:
                     self._cache[cache_key] = data
                     return data
                     
-            except aiohttp.ClientError as e:
+            except (aiohttp.ClientError, AttributeError) as e:
                 last_error = e
+                # If connector/session was closed, try to recreate it
+                if "Connector is closed" in str(e) or isinstance(e, AttributeError):
+                    logger.warning(f"Session issue detected, recreating session: {e}")
+                    self.session = None
+                    await self._ensure_session()
+                
                 if attempt < max_retries - 1:
                     logger.warning(f"Request failed (attempt {attempt + 1}): {e}, retrying in {delay}s")
                     await asyncio.sleep(delay)