ARM64 - Minor regression from optimizing 'a % b' #67983
Description
As a result of this PR: #66407 there is a minor regression where 'a / b' is not CSE'ed with cases of 'a % b' as the optimization removes the 'SubMulDiv' optimization when the constant is the power of 2.
This isn't a trivial problem and not an urgent one. The PR's optimization for 'a % b' is better to have than CSEs between 'a % b' and 'a / b'.
Marking this as future for now.