[llvm] [VPlan] Don't fold live ins with both scalar and vector operands (PR #154067)

[Florian Hahn via llvm-commits]

================

----------------
fhahn wrote:

Ok that's strange, it worked as expected for me when I changed it to have trip count of 4 with VF=4. The wide canonical IV is created if the trip count is < VF, as then it tries to fold the tail instead of creating an un-excutable vector loop

https://github.com/llvm/llvm-project/pull/154067