GitLab

Restructured how force and energy computations are done to make it easier to write plugins for the CUDA platform

Fixed getTime() calls in _openMMBrookInterface()
Based on Sherm's suggestions:
changed '#ifdef WIN32' to '#ifdef WIN32 _MSC_VER;
reordered include files so that system files after OpenMM files
all include files should be C++ files now (as opposed to C files (<math.h> -> <cmath>

License header changes; force duplication factor to 4; fixed potential problem w/ Shake -- if no constraints, pgm crashes

Made setting of RNG seed consistent w/ logic used for Cuda 

Minor edits; force duplicationFactor to 1 if no. particles < 65; for system w/ 32 atoms, NAN were appearing in nonbonded force term; plan to fix later

Added more ik,jl gathers for bonded forces
Redid shared #define setting in CMakeList.txt

Fixed setting of runtime in BrookPlatform() constructor (constructor w/o args):
if 'brt_runtime' env variable not set, then
internal 'runtime' variable was NULL -- this led to a fault when _setBrookRuntime() was called

Commented out static platform variable, which may have caused initialization problems
Some of the Brook tests now succeed on my computer for the first time.

Added license to kernel and kernel include files