[llvm] [MCP] Move dependencies if they block copy propagation (PR #105562)

[Quentin Colombet via llvm-commits]

qcolombet wrote:

> I think I could inline the lambda in the BFS shall I do it?

I'm not sure what this would look like, but generally speaking I like subroutines (I even prefer static functions to lambdas) because it makes the contract between the functionality and the surrounding code more obvious and forces you to document the API (what the input arguments are and what is returned mean!)

https://github.com/llvm/llvm-project/pull/105562