qa: file system deletion not complete because starter fs already destroyed
During cleanup of ceph, the file systems are not cleaned up which causes unnecessary MDS failover messages:
2021-02-26T13:19:02.521 DEBUG:tasks.cephfs.filesystem:already dead... ... 2021-02-26T13:19:46.979 INFO:teuthology.orchestra.run.smithi138.stdout:2021-02-26T13:19:18.871401+0000 mon.a (mon.0) 498 : cluster [WRN] Replacing daemon mds.c as rank 0 with standby daemon mds.d
The "starter" file system was already deleted by the cephfs task runner so when the ceph task tries to clean that up, it says the fs is already dead.
Make the cephfs task runner delete all file systems during tearDown.