cuda-embed-ptx macro and cuda architectures #140
Description
using owl in a project that also uses cuda for other stuff creates fatal build errors if that project sets cuda architectures for its cuda sources. to fix this we have to somehow disable the cuda architecture settings (only) within that macro, and leave them for other, regular cuda files.
right now only workaround is to not use cuda architectures at all in the main project (which is a bummer, and major inconvenience)