fix port (#3863)

Author: yangjianfengo1

tests/layers/test_moba_attention.py

@@ -359,7 +359,7 @@ def test_server(self):
             model=model_path,
             tensor_parallel_size=2,
             max_model_len=131072,
-            engine_worker_queue_port=8793,
+            engine_worker_queue_port=int(os.getenv("FD_ENGINE_QUEUE_PORT")),
             max_num_seqs=32,
             quantization="wint4",
             enable_chunked_prefill=True,