[input validation] pcs shall check that location constraint always relates to some nodes in one way or another

[jnpkrn]

# pcs constraint location dummy rule id=till-the-new-year score=-INFINITY date gt 2020-01-01 
# echo $?
> 0

->

      <rsc_location id="location-dummy" rsc="dummy">
        <rule id="till-the-new-year" score="-INFINITY">
          <date_expression id="till-the-new-year-expr" operation="gt" start="2020-01-01"/>
        </rule>
      </rsc_location>

This is a degenerate location constraint that achieves exactly nothing
but more complex overall configuration -- there's no actual "location"
specified in there.

At the very least, this shall raise a warning.

Relevant:
https://bugs.clusterlabs.org/show_bug.cgi?id=5415

[tomjelinek]

ack, tracked at https://bugzilla.redhat.com/show_bug.cgi?id=1791791