Better handling of floating types #1
Description
As of now the library must be compiled for either float or double.
It would be great if we could instead provide the users with a library that works for either type.
This limitation comes from UAMMD, which shares the same "compile for a certain precision" quirk.
One easy solution would be to compile two libraries (a la FFTW) and write a small frontend that redirects to the correct version depending on the input type.
We would end with two versions of each library (say psef.so and psed.so).