Properly processing urey bradley terms between openmm and parmed

[ahy3nz]

trafficstars

Based on the current workflow, converting an mbuild Compound and foyer Forcefield to an OpenMM System does not distinguish HarmonicBondForce with AmoebaUreyBradleyForce (the openmm system lumps 1-3 bonds and 1-2 bonds into the same HarmonicBondForce). However, ParmEd is able to distinguish 1-3 bonds and 1-2 bonds in structure.bonds and structure.urey_bradleys when ParmEd creates a Structure from an OpenMM Topology+System.

Parmed loops through the OpenMM System.getForces(). If HarmonicBonds have already been processed, then parmed will process any subsequent HarmonicBonds as urey bradleys.

The current idea is to separate out the 1-3 and the 1-2 bonds into different HarmonicBondForces in the OpenMM system after this line