chore: default to pytest asyncio-mode=auto (#2730)

# What does this PR do?

previously, developers who ran `./scripts/unit-tests.sh` would get
`asyncio-mode=auto`, which meant `@pytest.mark.asyncio` and
`@pytest_asyncio.fixture` were redundent. developers who ran `pytest`
directly would get pytest's default (strict mode), would run into errors
leading them to add `@pytest.mark.asyncio` / `@pytest_asyncio.fixture`
to their code.

with this change -
- `asyncio_mode=auto` is included in `pyproject.toml` making behavior
consistent for all invocations of pytest
- removes all redundant `@pytest_asyncio.fixture` and
`@pytest.mark.asyncio`
 - for good measure, requires `pytest>=8.4` and `pytest-asyncio>=1.0`

## Test Plan

- `./scripts/unit-tests.sh`
- `uv run pytest tests/unit`

tests/unit/providers/utils/inference/test_openai_compat.py

@@ -4,7 +4,6 @@
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
 
-import pytest
 
 from llama_stack.apis.common.content_types import TextContentItem
 from llama_stack.apis.inference import (
@@ -23,7 +22,6 @@ from llama_stack.providers.utils.inference.openai_compat import (
 )
 
 
-@pytest.mark.asyncio
 async def test_convert_message_to_openai_dict():
     message = UserMessage(content=[TextContentItem(text="Hello, world!")], role="user")
     assert await convert_message_to_openai_dict(message) == {
@@ -33,7 +31,6 @@ async def test_convert_message_to_openai_dict():
 
 
 # Test convert_message_to_openai_dict with a tool call
-@pytest.mark.asyncio
 async def test_convert_message_to_openai_dict_with_tool_call():
     message = CompletionMessage(
         content="",
@@ -54,7 +51,6 @@ async def test_convert_message_to_openai_dict_with_tool_call():
     }
 
 
-@pytest.mark.asyncio
 async def test_convert_message_to_openai_dict_with_builtin_tool_call():
     message = CompletionMessage(
         content="",
@@ -80,7 +76,6 @@ async def test_convert_message_to_openai_dict_with_builtin_tool_call():
     }
 
 
-@pytest.mark.asyncio
 async def test_openai_messages_to_messages_with_content_str():
     openai_messages = [
         OpenAISystemMessageParam(content="system message"),
@@ -98,7 +93,6 @@ async def test_openai_messages_to_messages_with_content_str():
     assert llama_messages[2].content == "assistant message"
 
 
-@pytest.mark.asyncio
 async def test_openai_messages_to_messages_with_content_list():
     openai_messages = [
         OpenAISystemMessageParam(content=[OpenAIChatCompletionContentPartTextParam(text="system message")]),

tests/unit/providers/utils/memory/test_vector_store.py

@@ -13,7 +13,6 @@ from llama_stack.apis.tools import RAGDocument
 from llama_stack.providers.utils.memory.vector_store import content_from_data_and_mime_type, content_from_doc
 
 
-@pytest.mark.asyncio
 async def test_content_from_doc_with_url():
     """Test extracting content from RAGDocument with URL content."""
     mock_url = URL(uri="https://example.com")
@@ -33,7 +32,6 @@ async def test_content_from_doc_with_url():
         mock_instance.get.assert_called_once_with(mock_url.uri)
 
 
-@pytest.mark.asyncio
 async def test_content_from_doc_with_pdf_url():
     """Test extracting content from RAGDocument with URL pointing to a PDF."""
     mock_url = URL(uri="https://example.com/document.pdf")
@@ -58,7 +56,6 @@ async def test_content_from_doc_with_pdf_url():
         mock_parse_pdf.assert_called_once_with(b"PDF binary data")
 
 
-@pytest.mark.asyncio
 async def test_content_from_doc_with_data_url():
     """Test extracting content from RAGDocument with data URL content."""
     data_url = "data:text/plain;base64,SGVsbG8gV29ybGQ="  # "Hello World" base64 encoded
@@ -74,7 +71,6 @@ async def test_content_from_doc_with_data_url():
         mock_content_from_data.assert_called_once_with(data_url)
 
 
-@pytest.mark.asyncio
 async def test_content_from_doc_with_string():
     """Test extracting content from RAGDocument with string content."""
     content_string = "This is plain text content"
@@ -85,7 +81,6 @@ async def test_content_from_doc_with_string():
     assert result == content_string
 
 
-@pytest.mark.asyncio
 async def test_content_from_doc_with_string_url():
     """Test extracting content from RAGDocument with string URL content."""
     url_string = "https://example.com"
@@ -105,7 +100,6 @@ async def test_content_from_doc_with_string_url():
         mock_instance.get.assert_called_once_with(url_string)
 
 
-@pytest.mark.asyncio
 async def test_content_from_doc_with_string_pdf_url():
     """Test extracting content from RAGDocument with string URL pointing to a PDF."""
     url_string = "https://example.com/document.pdf"
@@ -130,7 +124,6 @@ async def test_content_from_doc_with_string_pdf_url():
         mock_parse_pdf.assert_called_once_with(b"PDF binary data")
 
 
-@pytest.mark.asyncio
 async def test_content_from_doc_with_interleaved_content():
     """Test extracting content from RAGDocument with InterleavedContent (the new case added in the commit)."""
     interleaved_content = [TextContentItem(text="First item"), TextContentItem(text="Second item")]

tests/unit/providers/utils/test_model_registry.py

@@ -87,18 +87,15 @@ def helper(known_provider_model: ProviderModelEntry, known_provider_model2: Prov
     return ModelRegistryHelper([known_provider_model, known_provider_model2])
 
 
-@pytest.mark.asyncio
 async def test_lookup_unknown_model(helper: ModelRegistryHelper, unknown_model: Model) -> None:
     assert helper.get_provider_model_id(unknown_model.model_id) is None
 
 
-@pytest.mark.asyncio
 async def test_register_unknown_provider_model(helper: ModelRegistryHelper, unknown_model: Model) -> None:
     with pytest.raises(ValueError):
         await helper.register_model(unknown_model)
 
 
-@pytest.mark.asyncio
 async def test_register_model(helper: ModelRegistryHelper, known_model: Model) -> None:
     model = Model(
         provider_id=known_model.provider_id,
@@ -110,7 +107,6 @@ async def test_register_model(helper: ModelRegistryHelper, known_model: Model) -
     assert helper.get_provider_model_id(model.model_id) == model.provider_resource_id
 
 
-@pytest.mark.asyncio
 async def test_register_model_from_alias(helper: ModelRegistryHelper, known_model: Model) -> None:
     model = Model(
         provider_id=known_model.provider_id,
@@ -122,13 +118,11 @@ async def test_register_model_from_alias(helper: ModelRegistryHelper, known_mode
     assert helper.get_provider_model_id(model.model_id) == known_model.provider_resource_id
 
 
-@pytest.mark.asyncio
 async def test_register_model_existing(helper: ModelRegistryHelper, known_model: Model) -> None:
     await helper.register_model(known_model)
     assert helper.get_provider_model_id(known_model.model_id) == known_model.provider_resource_id
 
 
-@pytest.mark.asyncio
 async def test_register_model_existing_different(
     helper: ModelRegistryHelper, known_model: Model, known_model2: Model
 ) -> None:
@@ -137,7 +131,6 @@ async def test_register_model_existing_different(
         await helper.register_model(known_model)
 
 
-@pytest.mark.asyncio
 async def test_unregister_model(helper: ModelRegistryHelper, known_model: Model) -> None:
     await helper.register_model(known_model)  # duplicate entry
     assert helper.get_provider_model_id(known_model.model_id) == known_model.provider_model_id
@@ -145,18 +138,15 @@ async def test_unregister_model(helper: ModelRegistryHelper, known_model: Model)
     assert helper.get_provider_model_id(known_model.model_id) is None
 
 
-@pytest.mark.asyncio
 async def test_unregister_unknown_model(helper: ModelRegistryHelper, unknown_model: Model) -> None:
     with pytest.raises(ValueError):
         await helper.unregister_model(unknown_model.model_id)
 
 
-@pytest.mark.asyncio
 async def test_register_model_during_init(helper: ModelRegistryHelper, known_model: Model) -> None:
     assert helper.get_provider_model_id(known_model.provider_resource_id) == known_model.provider_model_id
 
 
-@pytest.mark.asyncio
 async def test_unregister_model_during_init(helper: ModelRegistryHelper, known_model: Model) -> None:
     assert helper.get_provider_model_id(known_model.provider_resource_id) == known_model.provider_model_id
     await helper.unregister_model(known_model.provider_resource_id)

tests/unit/providers/utils/test_scheduler.py

@@ -11,7 +11,6 @@ import pytest
 from llama_stack.providers.utils.scheduler import JobStatus, Scheduler
 
 
-@pytest.mark.asyncio
 async def test_scheduler_unknown_backend():
     with pytest.raises(ValueError):
         Scheduler(backend="unknown")
@@ -26,7 +25,6 @@ async def wait_for_job_completed(sched: Scheduler, job_id: str) -> None:
     raise TimeoutError(f"Job {job_id} did not complete in time.")
 
 
-@pytest.mark.asyncio
 async def test_scheduler_naive():
     sched = Scheduler()
 
@@ -87,7 +85,6 @@ async def test_scheduler_naive():
     assert job.logs[0][0] < job.logs[1][0]
 
 
-@pytest.mark.asyncio
 async def test_scheduler_naive_handler_raises():
     sched = Scheduler()