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

[Pankaj Dwivedi via llvm-commits]

PankajDwivedi-25 wrote:

> the propagation behavior might be **the result will be divergent if a specific source operand is divergent or other complex rule**

Yes, it can be seen as this way as well, currently we try to prove it Uniform. The new approach you are suggesting will also look similar. At all the places will be trying to prove divergent instead of uniform in the current patch.

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