Atomtypes parameters needs some Unit love

[justinGilmer]

For our charges, etc. in an atomtype, we do not support setting the mass, charge, etc of an atomtype on an atomtype by atomtype basis. We support setting these in the upper level ffmetadata , but not past that.

[uppittu11]

So in the case that the ParamUnitsDef is a dimension, we should check this dict first: https://github.com/mosdef-hub/gmso/blob/master/gmso/utils/ff_utils.py#L105

Otherwise, the ParamUnitsDef should be parsed by Unyt as it is now.

[mattwthompson]

I think this issue warrants longer discussion at some point - we have been extremely careful with unit handling in any Potential class, but not as careful with attributes of other classes that have important physical meaning.