GitLab

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

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 uninitialized bits occurring in the PaddedNumAtoms regions by properly constructing them. This should hopefully get rid of some of the NaN errors we've been seeing.

OpenCLArray is no longer templatized and doesn't provide a host buffer.  This is in preparation for adding mixed/double precision support.

Workaround for Nvidia compiler bug where compute level 1.2 processors try to compile double precision code

On Fermi, store forces as 64 bit fixed point and update with atomic operations.  Also various other minor optimizations.