[llvm] [LAA] Avoid unprofitable unit-stride versioning (PR #225331)

[Florian Hahn via llvm-commits]

https://github.com/fhahn commented:

> [dtcxzyw/llvm-opt-benchmark-nightly#1477](https://github.com/dtcxzyw/llvm-opt-benchmark-nightly/pull/1477) seems to show a significant decrease in loop loads eliminated. Presumably this is coming from the LoopLoadElimination pass? It would be good to have some tests similar to test/Transforms/LoopLoadElim/unknown-stride-known-dep.ll that expose the cases this PR is changing.


I think LLE is somewhat sensitive to the generated predicates, would be good to double check all the loops changed are actually single-iteration

https://github.com/llvm/llvm-project/pull/225331