[llvm] [llvm][ARM] Add a cortex-m4f alignment hazard recognizer (PR #126991)

[David Green via llvm-commits]

https://github.com/davemgreen commented:

Thanks - how do you think the hazard recognition method is doing? Does it seem to be performing OK, and have ran performance results from it? (The results I have still look a bit wobbly).

The scheduling might not be doing the best at the moment as I believe we don't run T2->T1 size reduce until after it has past-ra scheduled. The results it comes up with might be a bit random.

https://github.com/llvm/llvm-project/pull/126991