Support model overrides for auto-review

[won-openai]

Why

Some parent models need Guardian auto-review to run on the active parent model instead of the provider-preferred review model, while older /models payloads still need to deserialize cleanly when they do not include the new catalog field.

What changed

• Added optional ModelInfo::auto_review_model_override metadata with omitted-field compatibility.
• Updated Guardian model selection so Some(true) uses the parent turn model while missing or false values keep the existing preferred-review-model path and parent fallback.
• Added focused Guardian coverage for override true and missing/false behavior, and updated ModelInfo constructors in local catalogs and fixtures for the new optional field.

Validation

• cargo test -p codex-protocol
• cargo test -p codex-core guardian_review_uses_ --lib
• just fmt
• just fix -p codex-core
• git diff --check origin/main

A full cargo test -p codex-core run compiled and progressed through the suite, then aborted with a stack overflow in thread_manager::tests::resume_and_fork_do_not_restore_thread_environments_from_rollout.

[won-openai]

@codex review

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 58f0201cf8

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".