Adding CUDA and openCL SSA (pysb#427)

Add new SSA simulators using CUDA (Nvidia GPU only) and openCL (CPU and 
multi-GPU). These simulators provide orders of magnitude higher throughout
compared to the BNG SSA and Stochkit simulators.

---------

Co-authored-by: Alex Lubbock <[email protected]>
Co-authored-by: martinapr <[email protected]>
Co-authored-by: Jeremy Muhlich <[email protected]>