Temporarily add "-Wno-deprecated-gpu-targets" to set_runtime

[vyasr]

As of CUDA 12.8, when compiling for sm75 nvcc prints a warning

nvcc warning : Support for offline compilation for architectures prior to '<compute/sm/lto>_75' will be removed in a future release (Use -Wno-deprecated-gpu-targets to suppress warning).

We could add the suppression flag to all of RAPIDS, but given that we have a centralized way of linking to the CUDA runtime anyway perhaps we should simply set the compile flag in rapids_cuda_set_runtime where we link to the CUDA runtime.

This change would have the nice side effect of immediately telling us if any RAPIDS project is not using rapids_cuda_set_runtime, which could lead to unrelated subtle bugs around linking the CUDA targets unless they do things just right.