Description of problem:

When on a node that was put into maintenance-mode pacemaker is shut down all resources keep active.
That being detected by sbd when the pacemaker-connection goes away makes sbd
think the shutdown was ungracefully and thus it expects pacemaker to reconnect
immediately (watchdog-timeout).


Version-Release number of selected component (if applicable):

sbd-1.4.0-8.el8

How reproducible:

100%

Steps to Reproduce:
1. Setup cluster using sbd
2. Start cluster and wait till some resources are started on the node you want
   to trigger the missbehaviour on
3. Bring the node into maintenance-mode using 
   'pcs node maintenance {your-test-node}'
4. Issue 'systemctl stop pacemaker' on {your-test-node}

Actual results:

Node is rebooted by sbd

Expected results:

Sbd should detect that the pacemaker-shutdown is still graceful and not trigger
a reboot.

Additional info:

The correction would be needed e.g. in an pacemaker-upgrade-scenario where the node is first set to maintenance. Then pacemaker is shut down, upgraded, restarted and the node brought back from maintenance-mode.
The scenario is delicate though as the node still running resources then without
pacemaker-control could still be watchdog-fenced - without any impact on the
node of course. Preventing a node in maintenance-mode from being watchdog-fenced 
sounds like a solution here.