[llvm] [LV]Reject narrow VFs for narrow FP reductions fed by gappy interleaves (PR #222422)

[Florian Hahn via llvm-commits]

https://github.com/fhahn commented:

> the load fetches at least twice the used data, wasting bandwidth the cost
model can't see.

Is this only an issue for FP reductions? If the issue is wasting of bandwidth through the loads, should the interleave group cost be adjusted?

https://github.com/llvm/llvm-project/pull/222422