test: drain review cleanup callbacks

[Chris0Jeky]

Refs #2578

Summary

• Make the ReviewDecisionRail and ReviewMain cleanup stacks drain every callback even when one cleanup throws.
• Preserve LIFO teardown order and surface one failure or aggregate multiple failures after cleanup completes.
• Add direct regression coverage for the throwing-cleanup path in both affected review specs.

Verification

• Targeted ReviewDecisionRail + ReviewMain tests: 47/47
• Frontend unit suite: 5,710 passed, 3 skipped across 370 files
• Frontend typecheck: passed
• File-scoped ESLint: passed
• Mutation check: immediate rethrow caused the new test to fail because the earlier cleanup was skipped; helper restored and suite passed.
• git diff --check: passed

Two coordinator review passes completed for failure-path correctness and scope/regression quality.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, add credits to your account and enable them for code reviews in your settings.