fix race condition issue

put AIAvailabilityEnum.ENABLED to the end of model conversion

Azaion.Inference/main_inference.pyx

@@ -21,10 +21,11 @@ cdef class CommandProcessor:
     def __init__(self, int zmq_port, str loader_zmq_host, int loader_zmq_port, str api_url):
         self.remote_handler = RemoteCommandHandler(zmq_port, self.on_command)
         self.inference_queue = Queue(maxsize=constants_inf.QUEUE_MAXSIZE)
-        self.remote_handler.start()
-        self.running = True
         self.loader_client = LoaderClient(loader_zmq_host, loader_zmq_port)
         self.inference = Inference(self.loader_client, self.remote_handler)
+        self.running = True
+        self.remote_handler.start()
+
 
     def start(self):
         while self.running: