[llvm] [DA] Rewrite BanerjeeMIV test with safe APInt interval arithmetic (PR #207662)

[Ryotaro Kasuga via llvm-commits]

kasuga-fj wrote:

Can we continue using SCEV as before instead of APInt? I'm not sure whether it can provide more precise result, but if I understand correctly, it doesn't seem like there's some special reason to intentionally restricting it to APInt. Or do I misunderstand something?

https://github.com/llvm/llvm-project/pull/207662