Scenario:
Upgrade with disabled auto rollback was attempted before the certificates were expired.

The upgrade failed.

After a few days of debugging a rollback was attempted.

The certificates have expired for the old cluster, so at this point we have to take execute steps to recover from expired certificates:
https://docs.openshift.com/container-platform/4.15/backup_and_restore/control_plane_backup_and_restore/disaster_recovery/scenario-3-expired-certs.html

Without it the rollback doesn't complete - stuck.

We should add a note that the control plane certificates on the original stateroot could have expired while the new release was running, and point to this recovery section