Maint: pacemaker.service: cleanup/commentary on TimeoutStopSec directive

Also drop the historical fallback reference to TimeoutSec, and unify
the pacemaker_remote.service counterpart, too.

The credit for insightful wordings belongs to Ken Gaillot (except for
mistakes incurred with my line-minimizing postprocessing that go on
my account).

[Original discussion touched the topic of incidence/relations between
TimeoutStopSec and "shutdown-escalation" cluster option (20 min by
default, possibly intentionally lower), but that's left aside for
now.]