[llvm] [TTI] Introduce getInstructionUniformity API for flexible uniformity analysis (PR #137639)

[Pankaj Dwivedi via llvm-commits]

PankajDwivedi-25 wrote:

> During the divergence propagation, query the hook before marking user instruction divergent. Within the hook, it will decide whether the `divergentOp` will make the result value divergent

Thanks for your review.

As per our design discussion, we want the uniformity decision to be made within the analysis pass itself rather than querying the target about it. For that purpose, during the initialization phase, we query the target and record the target-dependent uniformity for later divergence propagation.

This approach will be flexible to handle any complex target-dependent uniformity, both for IR and MIRs.

https://github.com/llvm/llvm-project/pull/137639