GitLab

Improve the performance of sending the AmoebaVdwLambda to Cuda using pinned host memory; Updated the AmoebaVdwForceProxy to version 3, and added backward compatibility to version 2; updated TestAPIUnits.py to handle the per particle lambda flag

This reverts commit 5ef91d22.

* Improve vec3

* Added a ``__neg__`` operator overload so that -Vec3(1, 2, 3) returns
  the same thing as Vec3(-1, -2, -3)
* Derived Vec3 from namedtuple instead of tuple. This allows you to
  access the 3 elements of the vector by name. i.e., vec.x, vec.y, vec.z

* Make sure we use floating point division all the time.

* Implementation of OPLS through DMS reader

* Improved Implementation of OPLS through DMS reader

* raise errors for invalid implicit solvent settings

- Test system is translated to the origin
- Energy is computed again after updating Gromacs input to reflect no-cutoff option (rcoulomb = rlist = rvdw = 0.0)