feat(chat): robust thread support (complex)

backend/src/inference.py

@@ -154,7 +154,7 @@ async def chat_completion(app,
     agent_run=None
 ):
     # TESTING OVERRIDE
-    return await mock_completion(app, agent_run)
+    # return await mock_completion(app, agent_run)
     db_model = (await app.get_model(model_id)).value
 
     model_engine = db_model['engine']