[llvm] [NVPTX] Remove Float register classes (PR #140487)

[Artem Belevich via llvm-commits]

https://github.com/Artem-B approved this pull request.

LGTM.

I've checked how far back ptxas supports FP operations on .b32/.b64 registers, and it appears to work in CUDA versions as old as 9.1: https://godbolt.org/z/nbvPe57dc


https://github.com/llvm/llvm-project/pull/140487