[llvm] [ValueTracking] Add support for `vector_reduce_{s, u}{min, max}` in `isKnownNonZero`/`computeKnownBits`. (PR #88169)

[via llvm-commits]

goldsteinn wrote:

If these changes are desirable ill add support for the rest of the reduce intrinsics.

https://github.com/llvm/llvm-project/pull/88169