[clang] [CUDA][Win32] Add `fma(long double,..)` to math forward declares. (PR #73756)

[Artem Belevich via cfe-commits]

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

LGTM in principle, once the decls carry a comment. While it may not fix the issue for all users, if it helps some, it's a move in the right direction. We can deal with other instances if/when we find them.

https://github.com/llvm/llvm-project/pull/73756