[llvm] MachineBlockPlacement: Add tolerance to comparisons (PR #67197)

[Matthias Braun via llvm-commits]

MatzeB wrote:

> IIUC, these functions do not satisfy weak ordering.  

...
> In that case, is it a good idea to expose these as pseudo-comparators?

Yes it seems this is only a strict partial order but not a weak ordering. But is this actually a problem, would people infer this from the function names and get confused if it isn't? Are you saying we should find a weak ordering here? I don't immediately see any benefits from that at least for my use case.

Or are you proposing that we should find a different name for these functions to set other expectations? But then what name should we chose, I personally didn't have expectations one way or another...

https://github.com/llvm/llvm-project/pull/67197