[llvm] [GlobalISel] An optimizing MIR builder (PR #132282)

[Amara Emerson via llvm-commits]

aemerson wrote:

I don't think this is necessary at all. We have combiners for doing anything more than basic constant folding. On top of that the comments in this PR could be more coherent.

Finally, I don't really think GIConstant is appropriate for long term usage. It should really be an extension of GConstant IMO and probably shouldn't exist, but that's not the main issue in this PR.

https://github.com/llvm/llvm-project/pull/132282