mgr/cephadm: updating maintenance health status in the serve loop

mgr/cephadm: updating maintenance health status in the serve loop

The HOST_IN_MAINTENANCE health warning would disappear when the ceph manager
restarted or failed over, even though hosts remained in maintenance mode.
This occurred because set_maintenance_healthcheck() was only called during
host state transitions, not consistently during the serve loop.

This commit changes the function name from set_maintenance_healthcheck() to update_maintenance_healthcheck()
and then adds a call to update_maintenance_healthcheck() in the serve loop
to ensure maintenance health warnings are restored and persist across
manager restarts, similar to how the global pause health warning is handled.

Fixes: https://tracker.ceph.com/issues/72034
Signed-off-by: Sam Goyal <sam.goyal@clyso.com>
fixed test

(cherry picked from commit de4d7a954a57420d02a320b8b2144edc7be81057)