GitLab

This change is intended to improve the behavior when multiple
CUDA contexts are created on a node with multiple CUDA GPUs,
each of which is set to process-exclusive mode.

Converted the routine for identifying molecules from a recursion to a loop.  This avoids stack overflows with very long linear molecules.

AmoebaMultipoleForce did not report particle groups, which could potentially lead to incorrect reordering

Integrators control when atom reordering gets done, instead of having it happen automatically when forces are evaluated.  This simplifies some code, and also improves unintuitive behavior with CustomIntegrator.

Fixes a minor windows compiler invocation issue where C:\Program Files becomes delimited by the space. Also enables debugging libraries to be built in debug mode for WIN32 + Visual Studio. 

Fixes nvcc terminals popping up during CUDA JIT compilation due to invoking system("nvcc ..."). Replaced with MSVC's native CreateProcess() function.