Feature/ab#32339 align ai scoring access and prompt controls

[jacobwillsmith]

Pull request overview

This pull request aligns the AI scoring and prompt tooling flow with the access-control and prompt-discipline model we want long term. It tightens server-side permission and feature gating for AI regenerate actions and prompt capture, standardizes the reviewer posture across the v1 prompts, and centralizes prompt-type identifiers so prompt execution and prompt capture do not drift apart.

It also updates the review/assessment UI flow around AI scoring so the action bar, assessment scores widget, and application details page all reflect the same feature/permission state. The net effect is a cleaner scoring experience for reviewers and a safer prompt/debugging workflow in development.

Changes:

• Aligned AI scoring flow and review actions across the application details page, assessment scores widget, and review list action bar
• Added/normalized server-side permission and feature checks for attachment summaries, application analysis, AI scoring, and prompt capture access
• Standardized v1 prompt posture and restored generic guardrails to keep outputs conservative without overfitting
• Centralized AI prompt-type identifiers and wired prompt capture permission checks to explicit ABP permission checking

[github-actions]

🧪 Unit Test Results (Parallel Execution)

📊 Summary

Result	Count
✅ Passed	561
❌ Failed	0
⚠️ Skipped	0

📄 HTML Reports

• Merged Tests (HTML): Included in artifacts
  Generated automatically by CI.

[github-actions]

🧪 Unit Test Results (Parallel Execution)

📊 Summary

Result	Count
✅ Passed	561
❌ Failed	0
⚠️ Skipped	0

📄 HTML Reports

• Merged Tests (HTML): Included in artifacts
  Generated automatically by CI.