[llvm] TTI: Fix special casing vectorization costs of saturating add/sub (PR #97463)

[Simon Pilgrim via llvm-commits]

RKSimon wrote:

Can we keep the existing cost estimate code, but instead of returning Cost just stash it as a optional<ExpansionCost> value and fallback to it if the ISD legality fails?

https://github.com/llvm/llvm-project/pull/97463