[Chore] Use Sonnet 5 for default code reviews

[roomote-community]

Opened on behalf of Matt Rubens. Follow up by mentioning @roomote, in the web UI, or in Discord.

What changed

• Changed every provider recommendation that used Claude Opus 5 for code review to the provider's Claude Sonnet 5 model with medium reasoning.
• Covered OpenRouter's Balanced and Quick turnaround presets, Vercel AI Gateway, Anthropic, OpenCode, Amazon Bedrock, and GitHub Copilot.
• Kept Claude Opus 5 as the planning recommendation for those providers.
• Added optional per-role reasoning levels to legacy provider recommendation mappings and focused coverage for every affected preset.

Why this change was made

Claude Sonnet 5 provides a less expensive, less overly critical default for code review while retaining Opus 5 for planning work where its additional capability remains useful.

Impact

Deployments that apply these provider recommendations will use Claude Sonnet 5 at medium reasoning for code review. Existing explicitly saved model settings are unchanged until recommendations are applied again.

[roomote-community]

No code issues found. See task

_{Reviewed 010d9db}