[llvm] [AArch64] Lower factor-of-2 interleaved stores to STNP (PR #177938)

[Tomer Shafir via llvm-commits]

tomershafir wrote:

The reason for limiting to st2 is that we encountered this specific case on internal workloads and observed a significant speedup from the optimization. Haven't measured st3/st4 yet, and the overhead there should be higher.

https://github.com/llvm/llvm-project/pull/177938