[llvm] [TTI][Vectorize] Migrate masked/gather-scatter/strided/expand-compress costing (NFCI) (PR #165532)

[Shih-Po Hung via llvm-commits]

arcbbb wrote:

> What is the benefit of this approach, over having separate functions? It would seem that adding intrinsic-specific information would be simpler to pass to individual cost functions if needed.
> 

When I added the vp.load.ff cost hooks (#160470), I noticed vp_load, vp_load_ff, vp_gather, masked_load, and masked_expandload share almost the same interface.
it would be good to unify them behind a single interface to reduce boilerplate.

https://github.com/llvm/llvm-project/pull/165532