[clang] [ARM] Fix NaN behaviour for MVE compare intrinsics (PR #116371)

[David Green via cfe-commits]

https://github.com/davemgreen commented:

This seems to match what we do in the backend. LGTM

> The MVE intrinsics are defined as having the same behaviour as the instructions which they correspond to.

(They are defined to match the instructions they correspond to inside the current fp environment. i.e. they don't have side effects without trapping math and fast-math applies to them as required. That doesn't make a difference here though, they should still get the olt vs ult correct.)

https://github.com/llvm/llvm-project/pull/116371