[mlir][vector][llvm] Improve LLVM lowering of fmin/fmax reductions

[dcaballe]

We currently generate llvm reduce_fmin/reduce_fmax intrinsics whose semantics don't match the NaN handling of arith.fmin/arith.fmax operations in MLIR. We should generate llvm reduce_fminimum/reduce_fmaximum intrinsics instead, which have been recently added to LLVM.

[llvmbot]

@llvm/issue-subscribers-mlir-vectorops

[llvmbot]

@llvm/issue-subscribers-mlir-llvm

[unterumarmung]

https://reviews.llvm.org/D155869
https://reviews.llvm.org/D155877