Test: cts-scheduler: Drop per-resource fail counts

Add "#monitor_<interval>" to each per-resource fail count, where
<interval> is a non-promoted-role recurring monitor interval, if the
fail count is not INFINITY. Otherwise, add "#start_0"; it's more likely
that an INFINITY fail count came from a start or stop failure than from
1000000 monitor failures.

In cases where the resource with the fail count doesn't actually exist
in the CIB, append a dummy "#monitor_20000" string.

This is necessary after dropping support for per-resource fail counts.

These "#<action>_<interval>" suffixes wouldn't have occurred in CIBs
with these old feature sets, but adding them now is much easier than
recreating the tests with a current feature set.

For now, leave alone the "last-failure" nvpairs that don't have a
corresponding "fail-count" nvpair. Those last-failures should also have
an "#action_interval" suffix but so far its absence doesn't hurt
anything.

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>