[BugFix][Frontend] Use LoRA tokenizer in OpenAI APIs (#6227)

Co-authored-by: Cyrus Leung <cyrus.tl.leung@gmail.com>

tests/entrypoints/openai/test_serving_chat.py

@@ -38,5 +38,4 @@ async def _async_serving_chat_init():
 
 def test_async_serving_chat_init():
     serving_completion = asyncio.run(_async_serving_chat_init())
-    assert serving_completion.tokenizer is not None
-    assert serving_completion.tokenizer.chat_template == CHAT_TEMPLATE
+    assert serving_completion.chat_template == CHAT_TEMPLATE