[llvm] [AMDGPU][GlobalISel] Fold G_TRUNC(G_LSHR(x, 16)) into hi16 subregister copy in True16 mode (PR #196028)

[Matt Arsenault via llvm-commits]

arsenm wrote:

> > This should be done in tablegen (and I'd kind of expect we would have such a pattern already that works in sdag)
> 
> Yes, sdag already supported this pattern, when I implemented this PR: #192904, I noticed this issue.

But presumably in tablegen, so you should fix the pattern import 

https://github.com/llvm/llvm-project/pull/196028