[llvm] [MachineScheduler][ScheduleDAG] Add ability to bias scheduling by longest paths (PR #93223)

[Michael Maitland via llvm-commits]

michaelmaitland wrote:

> Then there's a bug in the patch, the compiler output should always be deterministic

It turns out that the non-determinism only existed in our downstream and was related to a different issue. That separate issue led me to believe that this patch was helping to improve scheduling. I suppose this patch helped that non-determinism show itself. After resolving the non-determinism in our downstream, the gains we saw "due to this patch" went away. As a result, I am closing this PR since it does not seem to help much.

https://github.com/llvm/llvm-project/pull/93223