Add recommend_reviewers_assign model selection feature setting

What does this MR do and why?

Gives the recommend reviewers foundational flow its own model selection feature setting, recommend_reviewers_assign. Today the flow sends duo_agent_platform, so it runs on whatever model "Agents & flows" uses, and the recommend_reviewers_assign default in the AI gateway is never read. With this change the flow gets its own "Recommend Reviewers" row in model selection and uses the AI gateway default, like the other foundational flows with their own setting such as resolve_dependency_bump and security_review.

A regular migration copies the duo_agent_platform feature setting to the new feature before the new code runs, so Duo Self-Hosted instances keep their own model, or keep the flow disabled if they disabled it. Like duo_developer, the new feature is also treated as a DAP feature for Duo Self-Hosted. GitLab.com and self-managed selections are not copied, so the flow uses the AI gateway default until an owner picks a model for it.

References

• AI gateway change that sets the default model: gitlab-org/modelops/applied-ml/code-suggestions/ai-assist!7181 (merged)

Screenshots or screen recordings

Before	After

MR acceptance checklist

Evaluate this MR against the MR acceptance checklist. It helps you analyze changes to reduce risks in quality, performance, reliability, security, and maintainability.