The idea here is to move towards multicolvar CVs such as the Q4 parameter
for those water molecules that are forming 4 hydrogen bonds.  We will have
that the input to any multicolvar can itself be a multicolvar which would allow
things like this to be computed using a multicolvar filter