[llvm] [RISCV][CodeGen][TTI]Use processShuffleMasks for cost estimations/actual per-register shuffles (PR #118103)

[Philip Reames via llvm-commits]

https://github.com/preames requested changes to this pull request.

There are at least three changes here which need to be separated.
1) Change to generic processShuffleMask, and tests for existing targets.  Is this is a generally beneficial change?  That needs to be justified, and reviewed.
2) Change to RISCV lowering to use processShuffleMask.  In particular, this needs significantly more testing!
3) Change to RISCV cost model.

I have no opinion on the order of 1 and 2, but 3 definitely needs to follow both 1 and 2.  

https://github.com/llvm/llvm-project/pull/118103