[llvm] [VPlan] Remove no longer needed VP intrinsic handling in VPWidenIntrinsicRecipe::computeCost. NFCI (PR #137573)

[Luke Lau via llvm-commits]

lukel97 wrote:

> LGTM. There is an unrelated question: is there a possibility that the operand is not of vector type?
> 
> ```
>   SmallVector<Type *> ParamTys;
>   for (unsigned I = 0; I != getNumOperands(); ++I)
>     ParamTys.push_back(
>         toVectorTy(Ctx.Types.inferScalarType(getOperand(I)), VF));
> ```

Yes I think so. I'm not sure why this isn't an issue at the moment. But we should be checking `onlyFirstLaneUsed` here probably.

https://github.com/llvm/llvm-project/pull/137573