Currently the backup/restore test relies on ssh and cluster-backup.sh to create the snapshots.

We should update it to create a one-off backup CRD instead and wait for the backup to finish. 

Note: this requires the graduation of the API, as we want to remove the TechPreview flag from the test case when we GA.

AC: 

• Use the one-off backup feature instead of ssh-based cluster-backup.sh
• Tests are still passing reliably
• Remove the TechPreview flag from the recovery test workflow 
  • Revert below two PRs: 
    • https://github.com/openshift/release/pull/41523
    • https://github.com/openshift/release/pull/41924