[llvm] [VPlan] Reassociate (x & y) & z -> x & (y & z) (PR #155383)

[Ramkumar Ramachandra via llvm-commits]

https://github.com/artagnon commented:

I think this makes sense. With my logical-op simplifications, which should go after this, we might get even more simplifications.

https://github.com/llvm/llvm-project/pull/155383