[llvm] [llvm-ir2vec] Automating ir2vec python bindings testing in llvm CI (PR #194593)

[Nishant Sachdeva via llvm-commits]

nishant-sachdeva wrote:

@boomanaiden154 

The `llvm-ir2vec` tests already run under this premerge job today ( downstream from check-llvm ) . Enabling the Python bindings would add just a few more unit tests to that existing run, keeping the whole IR2Vec ecosystem tested together in one place. 

The environment to build the bindings is also already present here (nanobind is available via MLIR_ENABLE_BINDINGS_PYTHON=ON, so the headers are present). 

I could move this to the ml-opt bots, but since this is still under development, catching regressions before they land would be better. 

Are there any specific concerns/criteria to put this along with the `ml-opt..` post merge checks ? 

https://github.com/llvm/llvm-project/pull/194593