Adjust prometheus textcollector condition

Currently our textcollector can sometimes fail with 'Failed to
create/acquire the lock /home/backup/$server/lock.exclusive (timeout)."
Instead of checking on a borg lock file, check if our backup snapshot
dir exists which the backup script creates and removes. This should give
less false positives then our current method.

roles/prometheus_exporters/templates/prometheus-borg-textcollector.service.j2

 [Unit]
 Description=Prometheus Borg Exporter TextCollector
 After=network.target
-ConditionPathExistsGlob=!/root/.cache/borg/*/lock.roster
+# If the backup snapshot path exists, the backup script runs and a borg
+# lockfile exists and we can't run borg list
+ConditionPathExists=!/backup
 
 [Service]
 Type=oneshot