[flang-commits] [flang] [flang][FIRToSCF] Keep no-wrap flags when narrowing a typed induction variable (PR #219281)

[via flang-commits]

khaki3 wrote:

The reason of this change is the unrolling behavior. My previous PR changed it, and we see a slight calculation error, which this PR fixes it. I'll investigate more to support negative bounds while keep the previous unrolling behavior.

https://github.com/llvm/llvm-project/pull/219281