[llvm] [NVPTX] Add folding for cvt.rn.bf16x2.f32 (PR #116109)

[Tom Natan via llvm-commits]

tomnatan30 wrote:

This change is breaking triton tests (results in huge numeric disparities, e.g. https://github.com/triton-lang/triton/blob/main/python/test/unit/language/test_core.py), we'll need to revert until a fix forward can be merged.

https://github.com/llvm/llvm-project/pull/116109