[llvm] [AMDGPU] Fix missing `IsExact` flag when expanding vector binary operator (PR #86712)

[Matt Arsenault via llvm-commits]

arsenm wrote:

> I'll update the test case in #83155 once that patch is landed to test this patch.

The code is already tested, just add a case with exact 

https://github.com/llvm/llvm-project/pull/86712