fix(vlm): default affected GPT tool calls to no reasoning

[qin-ptr]

Description

Session commits using gpt-6-luna can fail when Chat Completions receives function tools with the automatically selected reasoning_effort: "low".

For the same gpt-6-luna request with tools and no explicit reasoning setting, the default changes from reasoning_effort: "low" to reasoning_effort: "none". This applies to GPT-5.4/5.5/5.6 and GPT-6 Sol/Luna aliases and snapshots. Requests without tools and the openai-codex Responses path retain their existing defaults. Explicit reasoning_effort and extra_request_body settings retain their existing precedence.

Human Involvement

• A human participated in the implementation or review loop
• This PR was generated entirely by AI agents without human participation in the loop

Related Issue

Fixes #5757

Related: #5762

Type of Change

• Bug fix (non-breaking change that fixes an issue)
• New feature (non-breaking change that adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Documentation update
• Refactoring (no functional changes)
• Performance improvement
• Test update

Changes Made

• Apply shared completion defaults after the text and vision builders attach tools.
• Select none by default for affected Chat Completions tool requests, excluding openai-codex.
• Add three lines of logic and move two existing calls; no new configuration or tests.

Testing

• Existing reasoning, extra-body, Codex, response-format, and thinking tests: 112 passed, 1 pre-existing failure. test_openai_async_completion_with_empty_tools_preserves_usage also fails against the unmodified backend because its expected usage dictionary omits completion_tokens_details.

• Offline A/B check confirms the default gpt-6-luna tool request changes from low to none.

• HTTP mock transport verified 56 actual OpenAI SDK request bodies across synchronous/asynchronous text and vision calls, including explicit configuration and unaffected-model controls. Codex Responses controls retained low.

• Ruff lint/format checks and git diff --check passed.

• I have added tests that prove my fix is effective or that my feature works

• New and existing unit tests pass locally with my changes

• I have tested this on the following platforms:

  • Linux
  • macOS
  • Windows

Checklist

• My code follows the project's coding style
• I have performed a self-review of my code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• Any dependent changes have been merged and published

Screenshots (if applicable)

Not applicable.

Additional Notes

This changes automatic defaults only. An explicitly configured non-none effort can still be rejected by an affected Chat Completions endpoint. GPT-6 Astra and GPT-6.1 Sol require Responses for tool calling and are outside this fix.