revert openai_compat changes and use OpenAIMixin for openai_chat_completion

tests/integration/inference/test_openai_completion.py

@@ -13,6 +13,13 @@ import pytest
 from ..test_cases.test_case import TestCase
 
 
+@pytest.fixture(autouse=True)
+def rate_limit_delay():
+    """Add delay between tests to avoid rate limiting from providers like Fireworks"""
+    yield
+    time.sleep(30)  # 30 second delay after each test
+
+
 def _normalize_text(text: str) -> str:
     """
     Normalize Unicode text by removing diacritical marks for comparison.