cephadm: add some coarse-grained locking

cephadm: add some coarse-grained locking

Take per-cluster 'deploy' lock for

 bootstrap
 deploy
 adopt
 rm-daemon
 rm-cluster

No locking for

 version
 pull
 run (this is never actually used anyway)
 shell
 enter
 ceph-volume
 unit (this is a clean pass-through to systemctl)
 logs
 ls
 check-host

The only one I'm not sure about is ceph-volume.  I don't want the c-v
list/inventory commands to block, but perhaps the deploy ones should...
but perhaps not, since there's no reason we can't deploy multiple OSDs
on different devices at the same time.

Signed-off-by: Sage Weil <sage@redhat.com>