[llvm] [AArch64,TTI] Remove RealUse check for vector insert/extract costs. (PR #146526)

[Florian Hahn via llvm-commits]

fhahn wrote:

> Given the regressions it might be better to avoid the release to give us a chance to address it, but I do think this is better and should produce more meaningful numbers for SLP vectorization overall. LGTM

Thanks! I'd be inclined to merge it now and potentially revert or restrict to apple-cpus only in the upcoming release, if we encounter any issue. WDYT?

https://github.com/llvm/llvm-project/pull/146526