we could possibly have a fallback option to force-delete if you (manually) confirm that the cluster is gone