AB#32012 increase AI token fallbacks

[jacobwillsmith]

Pull request overview

This pull request increases the AI completion token limits used by the runtime fallbacks after we hit output caps during testing. The change raises the fallback limits for attachment summaries, application analysis, and application scoring so the runtime has more headroom when operation-specific config is missing or insufficient.

Changes:

• Increased the generic runtime completion-token fallback from 150 to 2000
• Increased the attachment summary fallback from 500 to 2000
• Increased the application analysis fallback from 2500 to 4000
• Increased the application scoring fallback from 5000 to 8000
• Kept the change scoped to runtime fallback values only

[github-actions]

🧪 Unit Test Results (Parallel Execution)

📊 Summary

Result	Count
✅ Passed	565
❌ Failed	0
⚠️ Skipped	0

📄 HTML Reports

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

[jacobwillsmith]

Just need these fallback tokens updated for redip testing. Got incomplete results back during testing.