Enable image generation continuation

[nm-openai]

Summary

• continue sampling once after a successful image_generation_call so workflows can produce requested final text or artifact metadata alongside the image
• add a stable default-on image_generation_continuation feature flag so callers can opt out
• include saved image metadata in model-visible history and cover both default and disabled behavior in tests

Original ticket: https://www.notion.so/openai/Allow-image_gen-workflows-to-provide-post-generation-text-or-artifact-metadata-35d8e50b62b0817cadcafa1056c6e41c?source=copy_link

Tests

• just write-config-schema
• just fmt
• cargo test -p codex-features image_generation_continuation_is_stable_and_enabled_by_default
• cargo test -p codex-core image_generation
• just fix -p codex-core
• just fix -p codex-features

[github-actions]

Thank you for your submission, we really appreciate it. Like many open-source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution. You can sign the CLA by just posting a Pull Request Comment same as the below format.

---

I have read the CLA Document and I hereby sign the CLA

---

_{You can retrigger this bot by commenting recheck in this Pull Request. Posted by the CLA Assistant Lite bot.}

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: b861e9ea81

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".

[chatgpt-codex-connector]

Escape revised_prompt before developer injection

revised_prompt is model/service output derived from the user's image request, but it is concatenated directly into a developer-role message. A prompt containing newlines or instructions is replayed on the continuation with developer priority, elevating untrusted text above the user. Store it as quoted/structured metadata or otherwise delimit/escape it.

Useful? React with 👍 / 👎.