fix: disable retries on kubernetes-apply deploy jobs to preserve logs

[vigneshrajsb]

Summary

• Kubernetes-apply deploy jobs used backoffLimit: 3 + restartPolicy: OnFailure, which caused K8s to delete the pod when BackoffLimitExceeded was reached
• This made logs unrecoverable — the JobMonitor couldn't fetch them, resulting in empty logs that skipped archival
• Failed deploy jobs showed up in the UI with a disabled logs button since there was nothing to display
• Changed to backoffLimit: 0 + restartPolicy: Never to align with native helm deploy and build jobs, which ensures the pod is preserved for log retrieval and archival

Test plan

• Deploy a service with a bad manifest to trigger a kubernetes-apply failure
• Verify the failed pod is preserved (not deleted by K8s)
• Verify logs are archived to object store after failure
• Verify the UI shows viewable logs for the failed deploy job