fix(ci): add xdist_group to health endpoint and watsonx tests for parallel stability

- test_health_liveliness_endpoint: add xdist_group('proxy_health') to prevent timeout
- test_watsonx_gpt_oss tests: add xdist_group('watsonx_heavy') to prevent mock interference

Co-authored-by: Ishaan Jaff <ishaan-jaff@users.noreply.github.com>

tests/test_litellm/llms/watsonx/test_watsonx.py

@@ -208,6 +208,7 @@ def test_watsonx_completion_regular_model_includes_model_id(
 
 
 @pytest.mark.asyncio
+@pytest.mark.xdist_group("watsonx_heavy")
 async def test_watsonx_gpt_oss_prompt_transformation(monkeypatch):
     """
     Test that gpt-oss-120b model transforms messages to proper format instead of simple concatenation.
@@ -391,6 +392,7 @@ async def test_watsonx_gpt_oss_prompt_transformation(monkeypatch):
 
 
 @pytest.mark.asyncio
+@pytest.mark.xdist_group("watsonx_heavy")
 async def test_watsonx_gpt_oss_uses_async_http_handler():
     """
     Test that verifies async HTTP client is used when fetching HuggingFace templates.

tests/test_litellm/proxy/health_endpoints/test_health_endpoints.py

@@ -374,6 +374,7 @@ def proxy_client(monkeypatch):
         yield client
 
 
+@pytest.mark.xdist_group("proxy_health")
 def test_health_liveliness_endpoint(proxy_client):
     """
     Test that /health/liveliness endpoint returns 200 OK with "I'm alive!" message.