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

[Matt Arsenault via llvm-commits]

arsenm wrote:

> This leads the test to pass some times and to fail other times. The benefit of using the test I add in this patch, is that it is deterministic.

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

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