[pve-devel] Fix shutdown order of ceph daemons.

[Wolfgang Link]

It is important that ceph stop after pveproxy.

If ceph stops to early and the shutdown is cluster wide the vm loses disks
and can not shutdown.

This can ends in fencing the node.

[PATCH V2]
PATCH: Make a new systemd target.
:: refactor: ensure that all possible storages used in PVE will be up.

PATCH: Fix shutdown order of ceph daemons.
:: add after pve-storage to ensure all storage's will stay up before pve-ha-lrm shutdown.

[PATCH V3]
PATCH: Make a new systemd target.
:: add ceph-mgr which is new in Luminous.

PATCH: Fix shutdown order of ceph daemons.
:: add Wants to ensure pve-store will started.