Allow 64bit floating point in (python) bindings #144
Description
As of now pygpufit has type checks and only allows for np.float32.
The REALdatatype in Gpufit/definitions.h allows to switch between 32bit single prec. and 64bit double prec. floating point in Gpufit (and Cpufit?).
So I guess it is possible to allow np.float64 in pygpufit or float/double in other java/matlab bindings.