Test performance of purging a stale subvolume when there are many PVs in the cluster

[travisn]

As discussed here, when the stale subvolumes are listed or deleted, there is an expensive query for all PVs in the cluster. If there are many PVs in the cluster, this could cause the command to either take a long time, or even fail. The PVs might not even be rbd or cephfs, they might be from some other provider.

Let's test the following scenarios:

• How long to list the stale subvolumes when there are ~100 PVs
• How long to delete a stale subvolume when there are ~100 PVs
• How long to list the stale subvolumes when there are ~1000 PVs
• How long to delete a stale subvolume when there are ~1000 PVs

@yati1998 ptal