Ensure correct logic & initial bench

based on [00a4240](https://github.com/nuke-web3/sp1-poseidon2-testing/commit/00a424003737a64cecf29fec39cddc34b2a29c38):
```
$ cargo r -r -- --execute

Program executed successfully.
zkVM -> hash: 0x1e14b899468e0d5a12b454bda89d49185d39b39da9baaff3a052d69a73d73b8e
Number of cycles: 5_482_078
```

We should test:
- [ ] basic hash matches known good example outside included tests on upstream poseidon2 patch
- [ ] merkle proof matches known good example
- [ ] logic mirrors/mock what we actually need for https://github.com/celestiaorg/eq-service/pull/87

Inspect commit 00a4240 and first run `cargo r -r -- --execute` to establish the current result. Compare the basic hash and Merkle proof with known-good examples, then check whether the logic mirrors the needs described in eq-service PR #87. Done means those checks are covered and the initial benchmark is recorded.

Written by the indexing model from the issue text.