'atommasses' array doubled for NBO calculations

[TimoSommer]

For Gaussian singlepoint calculations with NBO appended, when parsing the .log file, I would expect the atommasses to stay just the same as for a singlepoint calculation. However, after the NBO calculation, the atommasses array suddenly is twice as long. Apparently, the second calculation appends the entire atomic masses once again to the atommasses array.
For example, if for a H2O singlepoint calculation the atomic masses would be (roughly)
[1, 1, 16]
then after the NBO calculation, the atommasses would now be
[1, 1, 16, 1, 1, 16]

Versions:
cclib: 1.8

[berquist]

A reason we will want to wait on this one is that we will get proper multi-parser support in version 2.0 on the main branch. There was a fix for this in #1076 that may work for the master branch that could be revisited, but almost all of our effort is going into the upcoming 2.0 release.