feat(tests): Enforce EIP-7928 ordering rules

[fselmo]

🗒️ Description

To be clear, the current tests are not broken because the spec is implemented correctly. However, we were not enforcing certain validation rules that are required by EIP-7928. These changes introduce an initial up-front check on the entire order of the BAL coming from t8n as defined in EIP-7928:

Ordering and Determinism

The following ordering rules MUST apply:

- Addresses: lexicographic (bytewise).
- Storage keys: lexicographic within each account.
- Block access indices: ascending within each change list.

BlockAccessList (t8n model)

We fully validate all ordering upfront, based on the above rules, for the entire BAL coming from t8n. This is crucial that it meets all the requirements for the EIP implementation.

BlockAccessListExpectation (test expectation definition class)

Address Ordering

We define the expectation as a dict with address: BalAccountChanges. We don't need to check that our ordering is correct here because we will validate that the BAL coming from t8n is already in the correct order. So if the addresses we define are present at all, they will be in the correct order already, and we are free to define this as an unordered dict in the expectation test class. This is especially "nice" because utilities like pre.deploy_contract() and pre.fund_eoa() will generate addresses for us and it would be very difficult to put the burden on the test writer to place these within an OrderedDict, for example.

Transaction Index and Slot Ordering

We DO validate that the transaction indexes and slots appear in the correct order, but only as a sub-sequence of the full list - there’s no need to define every element. Since these are defined within a Python list, we do impose on the test writer that the sub-sequence they define here should be in order to maintain readability and reduce confusion for a reader of the test. I don't think this is too much to ask nor too big of a burden to place.

✅ Checklist

• All: Ran fast tox checks to avoid unnecessary CI fails, see also Code Standards and Enabling Pre-commit Checks:

  uvx --with=tox-uv tox -e lint,typecheck,spellcheck,markdownlint

• All: PR title adheres to the repo standard - it will be used as the squash commit message and should start type(scope):.
• All: Considered adding an entry to CHANGELOG.md.
• All: Considered updating the online docs in the ./docs/ directory.
• All: Set appropriate labels for the changes (only maintainers can apply labels).

[fselmo]

The last commit here attempts to fix issues relate to consume direct with geth which we agreed, for now (and for consistency I think it's also good), on a chat would be hex-encoded integers.

Unrelated but maybe relevant, the current state of the tests on geth branch jwasinger/bal-execution (the latest BAL branch):

• Pass all consume rlp tests
• Fail for consume engine. But after debugging issues with the current engine flow in geth, pass all tests except for invalid tests (which is just an exception mismatch issue and can be mitigated via exception mapper) if / when changes similar to what are in this PR are applied.
• Fail consume direct which geth expects as uint16 currently but will be expecting hex-encoded integers which this PR fixes. I tested the current implementation changing those values to int and all consume direct tests pass so if we agree both on the EEST side and geth side, they should end up passing once the agreement is met (hex-encoded).

^ This comment is mostly meant as a sanity check on the current state of the tests in case it helps to review with this context.

[spencer-tb]

Some small comments for now! The validation functions look good to me. Will review the tests tmo :D

[fselmo]

Some small comments for now!

Sounds good, ty! I will address these today.

Will review the tests tmo

The tests here are mostly sanity checking unit tests (no actual BAL tests added here) to make sure things don't get moved around and break these expectations in the future. But yes it would be good to get a sanity check on the sanity check! 😄

---

edit: Suggestions applied in latest commit 🙏🏼

[spencer-tb]

LGTM! Thanks :)