[itzmekhokan]

Summary

• Declare [text, image] input modalities for gpt-image-* and chatgpt-image-* models so refinement prompts can resolve a capable model instead of failing with "No models found that support image_generation".
• Route prompts containing a reference image to OpenAI's /images/edits endpoint with multipart form data, while keeping /images/generations for text-only prompts.
• Split DALL-E 2 metadata to support image editing separately from DALL-E 3 generation-only models.

Fixes https://core.trac.wordpress.org/ticket/64918

Test plan

• Connect OpenAI in Settings > Connectors with a valid API key
• Enable the AI plugin image generation experiment
• Generate an image in an Image block using OpenAI
• Click Refine Image and submit a refinement prompt
• Confirm the image is updated without the "No models found that support image_generation" error
• Confirm text-only image generation still works
• Run composer lint (PHPCS + PHPStan)

[github-actions]

The following accounts have interacted with this PR and/or linked issues. I will continue to update these lists as activity occurs. You can also manually ask me to refresh this list by adding the props-bot label.

If you're merging code through a pull request on GitHub, copy and paste the following into the bottom of the merge commit message.

Co-authored-by: itzmekhokan <khokansardar@git.wordpress.org>

To understand the WordPress project's expectations around crediting contributors, please review the Contributor Attribution page in the Core Handbook.

[superdav42]

LGTM!
Thanks for your contribution.