refactor: split large test and source files into focused modules

[skamenan7]

What does this PR do?

Splits 3 Python files that exceeded 1000 lines into smaller, focused modules. Pure refactoring with no behavior changes. All public names remain importable from the same paths.

Original file	Lines	Split into
tests/unit/providers/utils/inference/test_openai_mixin.py	1438	4 files by concern
tests/unit/providers/responses/builtin/test_openai_responses.py	3226	7 files by feature group
src/llama_stack/providers/inline/scoring/basic/utils/ifeval_utils.py	3319	5 files by responsibility

test_openai_mixin.py — split by test concern:

• openai_mixin_helpers.py — shared impl classes (180L)
• conftest.py — shared fixtures, auto-discovered by pytest (16L)
• test_openai_mixin_models.py — model listing, registration, availability (580L)
• test_openai_mixin_provider_data.py — provider data API key, custom model lists (200L)
• test_openai_mixin_inference.py — inference enforcement, stream options, param passthrough (520L)

test_openai_responses.py — split by feature group:

• conftest.py — 11 shared fixtures (116L)
• test_openai_responses_helpers.py — shared async generator (47L)
• test_openai_responses_core.py — core response creation, instructions, store logic (857L)
• test_openai_responses_agent.py — failed streams, agent loop, incomplete details (500L)
• test_openai_responses_tools.py — tool calls, MCP, file search, strict field (670L)
• test_openai_responses_prompts.py — prompt template variants (579L)
• test_openai_responses_params.py — param passthrough, service tier, stream options (856L)

ifeval_utils.py — split by responsibility:

• ifeval_word_list.py — WORD_LIST data literal only, grandfathered as pure data (1552L)
• ifeval_support.py — language constants, regex patterns, utility functions (296L)
• ifeval_checkers_core.py — Instruction base class + 15 checker classes (828L)
• ifeval_checkers_format.py — 15 checker classes (703L)
• ifeval_utils.py — thin entry point: re-exports INSTRUCTION_DICT and INSTRUCTION_LIST (96L)

The external import in ifeval_scoring_fn.py is unchanged. Shared fixtures previously loaded via a root conftest.py plugin import are now in a directory-level conftest.py and auto-discovered by pytest.

Test Plan

uv run pytest tests/unit/providers/utils/inference/test_openai_mixin_models.py \
  tests/unit/providers/utils/inference/test_openai_mixin_provider_data.py \
  tests/unit/providers/utils/inference/test_openai_mixin_inference.py -v


uv run pytest tests/unit/providers/responses/builtin/ -v


uv run pre-commit run --all-files

[mergify]

This pull request has merge conflicts that must be resolved before it can be merged. @skamenan7 please rebase it. https://docs.github.com/en/pull-requests/collaborating-with-pull-requests/working-with-forks/syncing-a-fork

[skamenan7]

Hi @leseb , thanks for the PR# 5281 review and merge. this one follows that one for more changes. PTAL.