[clang] [llvm] [AMDGPU] Change CF intrinsics lowering to reconverge on predecessors (PR #108596)

[Nicolai Hähnle via cfe-commits]

nhaehnle wrote:

Somebody recently mentioned a test case that ended up producing hundreds of MBs or even GBs of code :)

Maybe the answer is that we don't support it and just report a compiler error. That may be a reasonable thing to do, but in any case having certainty about it would help the discussion here.

https://github.com/llvm/llvm-project/pull/108596