ui: ceph: add a health preflight to the cluster-wide restart

The MON, MGR and MDS bulk restart sent the request without any health
information and without a way to pass force. An operator only learned
that the cluster health blocks the restart from the resulting task
error. Show the checks that the backend reports as blocking before
asking for confirmation, and offer Force when every blocking check is
only a warning.

A failed health query has two causes. A 403 only means that this user
may not read the health while the backend still can, so the restart
runs as usual and force stays available. Any other failure means the
backend runs into the same problem, and then the cluster needs a look
first.

Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>