Implement utility functionality for energy comparison for parm@frosst

[davidlmobley]

Christopher Bayly is working towards building "parm@frosst-y" in SMIRFF format and needs some utility functionality to assist him as I describe below.

The context is that converting parm@frosst to SMIRFF format by hand would be quite the major endeavor (8000 parameter conversions at even only a couple of minutes each would take more than a month), so he's going to try to instead get a "reasonable chunk" of the easiest/most general chemistry done; we could perhaps call this "parm@frosst-y".

To assist him, he needs some utility functionality that will take a specified set of molecules which we already have parm@frosst types for and for which we can generate AMBER prmtop and crd files (such as the ZINC parm@frosst subset, or a separate MMFF94 set which is apparently also on CCL) and do single point energies on all of the molecules with parm@frosst and with his draft parm@frosst-y SMIRFF, breaking out energy differences by force type and, if possible, per atom index group, and (on the SMIRFF side) what the corresponding SMIRKS was, so that he can easily drill down and find out which molecules (and which fragments in those molecules) are leading to the largest energy differences.

I'll try to do this immediately since it will be a bottleneck for him otherwise (and I'll revisit #53 afterwards).

[davidlmobley]

Closed by #124 .