I am able to reproduce your issue. I have already informed our GKE team to investigate the issue further.
By the time being a workaround is to avoid the issue is to delete the claim corresponding PersistentVolume object and it will also delete the provisioned Compute Engine persistent disk [1], before you delete the cluster.