This makes the combination of Q6+LQ6+VOLUMES faster as you only
calculate the Q6 values in the region of interest and the surrounding
link cells.  In other words you avoid calculating things that you know
will make no contribution to the CV

[makedoc]

This change makes the driver less error prone

It requires a lot of regtests to be fixed,
mostly adding an mc file when required or replacing
COM with CENTER

Also if we run DFS on an adjacency matrix without an input colvar
then the values used in DFS clustering are the sums of the rows
of the adjacency matrix

This is an updated version of the regtest

If you had no flags then optional keywords were listed as compulsory.
this is now fixed [makedoc]

Added command to work out size of largest cluster including surface and fixed documentation for commit

[makedoc]

It is now possible to specify a different link cell cutoff for the
third bodies.  In addition, the way that connectivity is determined
in adjacency matrix actions works better given that we can now
operate with non-contact matrix contact matrices.  Also added some
cutoff stuff to HistogramBead.  This is currently not used in many
places

Added action to calculate matrix that tells us whether or not pairs of atoms are hydrogen bonded to each other

related to #137

    Related to #175

Related to #175

updated example
[makedoc]

[makedoc]

[makedoc]

changelog, regtest and bug fixes

close #172