[llvm] [AMDGPU] Set gfx13 LDS to 192KB and model LDS size via subtarget features (PR #214513)

[Chinmay Deshpande via llvm-commits]

chinmaydd wrote:

Hi @mariusz-sikora-at-amd, I think it would be reasonable to go ahead with this PR in its current state and we can discuss and update the terminology in https://github.com/llvm/llvm-project/pull/215681 

https://github.com/llvm/llvm-project/pull/214513