Maybe a bug, report for CV used MTD

[Ross, Sun]

Dear developers,
I run MTD with COMBINE_COLVAR method with following input, I got a
wrong value of CV.

    &COLVAR
      &COMBINE_COLVAR
       &COLVAR
        &COORDINATION
          ATOMS_FROM  4
          KINDS_TO        H
            R0 [angstrom] 1.0
            NN 10
            ND 20
        &END COORDINATION
       &END COLVAR
        ...
      &END COMBINE_COLVAR
    &END COLVAR

However if I change ATOMS_TO    3 7 8 9 11 12 14 (all H atoms),
it gave me the right answer.

Other ways are also tested without combine_colvar method with
KINDS_TO        H    or
ATOMS_TO    3 7 8 9 11 12 14
they gave the same value

So, I think it maybe a bug for conbine_colvar with   ATOMS_FROM and
KINDS_TO.

Regards,
sun