fix: improve test isolation for parallel execution

Fixes test failures that occur during parallel test execution (pytest -n 4)
due to module reloading issues with conftest.py reloading litellm.

Changes:
- Add module reload fixtures to ensure fresh references after conftest reloads
- Use patch.object and string-based patches instead of direct attribute assignment
- Use class name comparison instead of isinstance for reloaded modules
- Handle case where litellm is missing from sys.modules during parallel runs
- Move stream consumption inside patch contexts to avoid real API calls
- Mock litellm.acompletion instead of low-level HTTP handlers
- Add skipif decorator for enterprise-only test classes

Affected test files:
- test_container_integration.py
- test_responses_background_cost.py
- test_huggingface_embedding_handler.py
- test_vertex_ai_rerank_integration.py
- test_volcengine_responses_transformation.py
- test_pillar_guardrails.py
- test_litellm_pre_call_utils.py
- test_proxy_server.py
- test_converse_transformation.py
- test_chat_completions_handler.py
- test_aresponses_api_with_mcp.py
- test_anthropic_experimental_pass_through_messages_handler.py

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

tests/test_litellm/llms/huggingface/embedding/test_huggingface_embedding_handler.py

@@ -1,3 +1,4 @@
+import importlib
 import json
 import os
 import sys
@@ -15,7 +16,22 @@ MOCK_EMBEDDING_RESPONSE = [[0.1, 0.2, 0.3, 0.4, 0.5]]
 
 
 @pytest.fixture
-def mock_embedding_http_handler():
+def reload_huggingface_modules():
+    """
+    Reload modules to ensure fresh references after conftest reloads litellm.
+    This ensures the HTTPHandler class being patched is the same one used by
+    the embedding handler during parallel test execution.
+    """
+    import litellm.llms.custom_httpx.http_handler as http_handler_module
+    import litellm.llms.huggingface.embedding.handler as hf_embedding_handler_module
+
+    importlib.reload(http_handler_module)
+    importlib.reload(hf_embedding_handler_module)
+    yield
+
+
+@pytest.fixture
+def mock_embedding_http_handler(reload_huggingface_modules):
     """Fixture to mock the HTTP handler for embedding tests"""
     with patch("litellm.llms.custom_httpx.http_handler.HTTPHandler.post") as mock_post:
         mock_response = MagicMock()
@@ -27,7 +43,7 @@ def mock_embedding_http_handler():
 
 
 @pytest.fixture
-def mock_embedding_async_http_handler():
+def mock_embedding_async_http_handler(reload_huggingface_modules):
     """Fixture to mock the async HTTP handler for embedding tests"""
     with patch("litellm.llms.custom_httpx.http_handler.AsyncHTTPHandler.post", new_callable=AsyncMock) as mock_post:
         mock_response = MagicMock()

tests/test_litellm/llms/vertex_ai/rerank/test_vertex_ai_rerank_integration.py

@@ -2,6 +2,7 @@
 Integration tests for Vertex AI rerank functionality.
 These tests demonstrate end-to-end usage of the Vertex AI rerank feature.
 """
+import importlib
 import os
 from unittest.mock import MagicMock, patch
 
@@ -13,7 +14,14 @@ from litellm.llms.vertex_ai.rerank.transformation import VertexAIRerankConfig
 
 class TestVertexAIRerankIntegration:
     def setup_method(self):
-        self.config = VertexAIRerankConfig()
+        # Reload modules to ensure fresh references after conftest reloads litellm.
+        # This ensures the class being patched is the same one used by the tests.
+        import litellm.llms.vertex_ai.rerank.transformation as rerank_transformation_module
+        importlib.reload(rerank_transformation_module)
+
+        # Re-import after reload to get the fresh class
+        from litellm.llms.vertex_ai.rerank.transformation import VertexAIRerankConfig as FreshConfig
+        self.config = FreshConfig()
         self.model = "semantic-ranker-default@latest"
 
     @patch('litellm.llms.vertex_ai.rerank.transformation.VertexAIRerankConfig._ensure_access_token')

tests/test_litellm/responses/mcp/test_chat_completions_handler.py

@@ -625,7 +625,7 @@ async def test_acompletion_with_mcp_streaming_metadata_in_correct_chunks(monkeyp
             ],
         ),  # Final chunk with tool_calls
     ]
-    
+
     follow_up_chunks = [
         create_chunk("Hello"),
         create_chunk(" world", finish_reason="stop"),