[llvm] [VPlan] Remove non-reductions after simplifications. (PR #176795)

[via llvm-commits]

================
@@ -686,7 +686,6 @@ define i1 @reduction_with_const_or(ptr %A, i8 %n) {
 ; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
 ; CHECK:       [[VECTOR_BODY]]:
 ; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i1> [ zeroinitializer, %[[VECTOR_PH]] ], [ splat (i1 true), %[[VECTOR_BODY]] ]
----------------
ayalz wrote:

Scalar loop has %or.red holding false for first iteration and true for all remaining iterations.
What if the test returns %or.red instead of %or.next - which is correctly folded to true?

https://github.com/llvm/llvm-project/pull/176795