SCHED-300: Delete wait-for-checks-job if helm release failed

[rdjjke]

Problem

If soperator-activechecks helm release failed, a new attempt fails as well because K8s job "wait-for-checks-job" already exists.

Solution

Add hook-failed and before-hook-creation values to this K8s job's "helm.sh/hook-delete-policy" annotation.

Testing

To test:

1. Create a new cluster with a wrong path to the image used by active checks.
2. Wait until soperator-activechecks HR failed
3. Redeploy this HR

Release Notes

Fixed a bug that caused repeated attempts of cluster creation to fail if the deployment of active checks failed during the first attempt.