[llvm] [AMDGPU][GISel] Do not scalarize uniform v2f16/v2bf16 operations (PR #222359)

[Petar Avramovic via llvm-commits]

petar-avramovic wrote:

ok so keeping v2bf16 in sgprs looks too long with that lowering, but think we should leave v2f16 as it was

https://github.com/llvm/llvm-project/pull/222359