Bug fixes in 4.1:

[#1594] Fortran: OpenMM_DoubleArray_set fails
[#1599] AmberPrmtopFile ignores rigidWater flag
[#1604] Context constructor throws exception
[#1606] AmberInpcrdFile returns velocities with wrong units
[#1612] Context.applyConstraints() does not work for OpenCL
[#1613] PDBFile ignores atoms with alternate locations
[#1616] AmberPrmtopFile with constraints=HAngles computes the wrong
distance for some constraints
[#1629] NaN forces with AMOEBA system created through openmm app
[#1634] Variable time step integrators do not report step size correctly
[#1659] simtk.openmm.app.amberinpcrdfile: getPositions(asNumpy=True)
doesn't return correctly-wrapped Quantity
[#1664] Multipole energy incorrect for PME caclulations
[#1665] AMOEBA multipole forces incorrect for non-cubic boxes
[#1666] The induced dipoles are nans for AMOEBA runs
[#1667] The AMOEBA Van der Waals forces and energy disagree w/ TINKER
values