[SandboxIR][Bench] Test RAUW

[vporpo]

No description provided.

[alinas]

LGTM with some notes.

For future patches, consider if you can generate IR that can be reused between benchmarks. For example, the IR in generateRAUWIR could be used instead of generateBBWalkIR. The concern here being future code explosion and keeping the focus on what's being tested.

You can also have one where you aggregate some of the API being tested in a single benchmark.

[vporpo]

consider if you can generate IR that can be reused between benchmarks.

We could reuse the same IR across a few tests, but in general tests require different IR, which is why the IR generation function is passed as a parameter to genIR().