[llvm] [BOLT] Optimize basic block loops to avoid n^2 loop (PR #156243)

[Jakub Beránek via llvm-commits]

Kobzol wrote:

Well, hard to say then. Tbh the main issue with BOLT that we have is not RSS or even the BOLT optimization time, but the fact that it doubles the binary size of the optimized artifacts, due to `-use-old-text` not working properly. But that's another topic.

I suppose that this PR can be closed, now that the change has landed in https://github.com/llvm/llvm-project/pull/160407?

https://github.com/llvm/llvm-project/pull/156243