[llvm] [VPlan] Make simplifyRecipe more like InstCombine (PR #212968)

[Luke Lau via llvm-commits]

================
@@ -16,19 +16,16 @@ define void @test1() #0 personality ptr @__CxxFrameHandler3 {
 ; CHECK-NEXT:    [[TMP0:%.*]] = catchswitch within none [label %[[CATCH:.*]]] unwind to caller
 ; CHECK:       [[CATCH]]:
 ; CHECK-NEXT:    [[TMP1:%.*]] = catchpad within [[TMP0]] [ptr null, i32 64, ptr null]
-; CHECK-NEXT:    br label %[[FOR_BODY:.*]]
-; CHECK:       [[FOR_BODY]]:
 ; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
 ; CHECK:       [[VECTOR_BODY]]:
-; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, %[[FOR_BODY]] ], [ [[INC:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP2:%.*]] = call <16 x double> @llvm.floor.v16f64(<16 x double> splat (double 1.000000e+00)) [ "funclet"(token [[TMP1]]) ]
-; CHECK-NEXT:    [[INC]] = add nuw i32 [[INDEX]], 16
+; CHECK-NEXT:    [[IV:%.*]] = phi i32 [ 0, %[[CATCH]] ], [ [[INC:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[CALL:%.*]] = call double @floor(double 1.000000e+00) #[[ATTR1:[0-9]+]] [ "funclet"(token [[TMP1]]) ]
----------------
lukel97 wrote:

We now remove the dead vectorized call after RAUW-ing it, so now `willGenerateVectors` returns false and we don't vectorize anymore.

https://github.com/llvm/llvm-project/pull/212968