Fix test set duplicate display and auto-deletion logic

- Fix UI duplicate test sets from polling conflicts
- Implement upsert logic to merge UI and backend state
- Preserve active test sets during processing
- Only auto-delete FAILED test sets after 1-hour cutoff
- Add type ignore comments for DynamoDB Table attributes
- Fix PyArrow compute.equal type checking issue

Resolves test set lifecycle management issues and ensures
proper error display with conservative cleanup policies.

Author: Taniya Mathur