[llvm] AMDGPU: Add tonearest and towardzero roundings for intrinsic llvm.fptrunc.round (PR #104486)

[Changpeng Fang via llvm-commits]

changpeng wrote:

And what to do on global-isel path? I guess something similar. 
It will end up with similar complexity as before.

https://github.com/llvm/llvm-project/pull/104486