[llvm] [AMDGPU][GlobalISel] Select s16 G_MERGE_VALUES into wider scalars (PR #207999)

[Petar Avramovic via llvm-commits]

petar-avramovic wrote:

there are complications, should no longer use LLT::scalar, first need to rebase and make sure LLT::integer or LLT::float are used

https://github.com/llvm/llvm-project/pull/207999