Multi-threaded compilation of GPU binaries

[Hoshinova]

Hi,

I noticed that when compiling the CUDA and OptiX binaries the cicc compiler always runs in single threaded mode.

Is it possible to use multi-threading when compiling the CUDA and OptiX binaries?

[Alaska]

I believe this PR adds multi-threading support for CUDA/OptiX kernel compilation:

WIP: Rework CUDA kernel build to use separate compilation