- Covers the openai-codex + supports_reasoning=False path now that
Codex routes through _models_dev_reasoning_efforts.
- Mirrors the existing false-suppression test for symmetry.
- 9/9 tests pass.
- Remove openai-codex special case that called github_model_reasoning_efforts()
- Codex now falls through to _models_dev_reasoning_efforts() (full efforts)
- GitHub/Copilot still use the GitHub helper (caps at high)
- Added regression tests for both behaviors