[llvm] [LoopVectorize] Vectorize select-cmp reduction pattern for increasing integer induction variable (PR #67812)

[Mel Chen via llvm-commits]

Mel-Chen wrote:

Rebased, please take a look.
Additionally, we have encountered a new case after switch vectorization was enabled 54b624c229852eb33c200bc24bf267a5e5f6be4d.
The same issue might also occur with AnyOf reduction, although we haven't encountered it yet.

https://github.com/llvm/llvm-project/pull/67812