Adding nonbonded fixes to a forcefield

[CalCraven]

Commonly, things like fixing specific pair interactions, termed a non-bonded fix, can be written into an engine to override the 1-4 mixing rules. This is most typical for CHARMM simulations and charmm/amber forcefields. I will post a link to some literature discussing the use of these. We need to try to implement some support for this.
https://pmc.ncbi.nlm.nih.gov/articles/PMC5874203/

Would probably look like a new section in the topology that could specify two atom types or classes as map keys and write out that section if it's included in the topology.