A report job should run after a failed test job so that it can publish diagnostic information, but the final deployment must still require successful tests. Where should the status override be applied?

1. Set if: failure() on the deployment job.
2. Keep if: success() on the report job with needs: test and expect it to run after test fails.
3. Permit report with always(), while retaining the test-success dependency for deployment.
4. Set continue-on-error: true for all test steps and remove deployment dependencies.