[Tests] Disable retries and use context manager for openai client (#7565)

tests/entrypoints/openai/test_basic.py

@@ -2,6 +2,7 @@ from http import HTTPStatus
 
 import openai
 import pytest
+import pytest_asyncio
 import requests
 
 from vllm.version import __version__ as VLLM_VERSION
@@ -28,9 +29,10 @@ def server():
         yield remote_server
 
 
-@pytest.fixture(scope="module")
-def client(server):
-    return server.get_async_client()
+@pytest_asyncio.fixture
+async def client(server):
+    async with server.get_async_client() as async_client:
+        yield async_client
 
 
 @pytest.mark.asyncio