GitLab

 Conflicts:
	wrappers/python/simtk/openmm/app/amberprmtopfile.py

In fixing the merge conflict, I went ahead and fixed up the switchDistance logic
to match what I did in CharmmPsfFile.

Deleted GBVIForce

Fix #1213 addInteractionGroup should accept `set` values as parameters

apparently SWIG typemaps maps Python's list but not set with std::set.
Instead, Python's `set` is mapped with `std::unordered_set`.

The original code of `swigInputBuilder.py` is kindly in a mess. Too much
lines in a single method `writeMethods` which causes debug becomes
difficult. Also there are a lot of warnings about
[PEP8](https://www.python.org/dev/peps/pep-0008/).

A refactoring in the future may be a good idea.

Residue templates can specify atoms by name instead of index

AMD docstring fix

Deploy requires AWS region

Deploy development docs to S3

Better docstrings

Changes necessary for VS2015

Added missing support for LocalCoordinatesSite

addExtraParticles() can use bonds to select positions

ForceField will not create duplicate constraints

Fixed a race condition in energy computation

Update appveyor.yml

Fix a handful of compiler warnings