Compare constraints evaluation on winterfell and plonky3

### What should be done?

Implement a test harness in order to test evaluations of generated winterfell and plonky3 files.

### How should it be done?

Compare evaluations from existing tests present under `air-script/src/test`.
Compare behaviour on both plausible traces and generated random ones.

### When is this task done?

When all tests are implemented and pass, and the tes harness is properly documented.

### Additional context

Currently, we expect generated winterfell and plonky3 files from the same air-script are equivalent, but nothing guarantees it past the AIR part of the compilation pipeline.

The goal is to ensure that the behaviour matches after their respective codegen part.

Look at existing tests under `air-script/src/test` to understand the AIR constraint generation. The task is to build a test harness that compares evaluations from Winterfell and Plonky3 outputs. Start by examining the codegen pipeline for both targets, then write tests that run on both plausible and random traces. Verify that the compiled constraints behave identically.

Scritto dal modello di indicizzazione a partire dal testo della issue.