[llvm] [InstCombine] Fold `(trunc X)` into `X & Mask` inside `decomposeBitTestICmp` (PR #171195)

[Tirthankar Mazumder via llvm-commits]

wermos wrote:

The changes in LoopVectorize are the same as those in the `memchr` and `strchr` files, it's truncating an `i64` to an `i32` and then doing an equality comparison. Once again, I eyeballed the mask constant and it looks correct to me.

With this final test update, all the tests _should_ pass.

https://github.com/llvm/llvm-project/pull/171195