Implement the --stale-openstack action to cleanup OpenStack resources
that are known to be stale because:
* instances or volumes have been created a long time ago
* instances or volumes with REMOVE-ME in the name
* nodes that are locked but did not get an instance
Only the instances and volumes known to belong to the teuthology cluster
because they have ownedby=IP in the name are considered, where IP is the
IP address of the machine running the teuthology clusters.