[llvm] [VPlan] First step towards VPlan cost modeling. (PR #67934)

[Shih-Po Hung via llvm-commits]

arcbbb wrote:

I am trying to mitigate the cost difference caused by removeDeadRecipes(), since legacy cost model still count them.
In my implementation https://github.com/arcbbb/llvm-project/commit/cba398c0aa2199423c9e89885f9dafb2cc125a1f, 
I collect all scalar costs which are not seen in previous visited recipes.

https://github.com/llvm/llvm-project/pull/67934