[polly] [Polly] Reject SCoPs with an escaping scalar defined in an empty-doma… (PR #213656)

[Michael Kruse via llvm-commits]

Meinersbur wrote:

I don't see why we would need to reject the entire SCoP. If the producing ScopStmt is purged because it is known to be never executed and only required because of structural reasons, it can be replaced with a undef (or better: poison) value.

https://github.com/llvm/llvm-project/pull/213656