refactor(test-fill): rebuild the Engine X drift check on typed fixture models

[spencer-tb]

Description

Rebuild engine_x_checks.py per the review feedback on #3219: fixtures were compared as black-box JSON dicts, the BAL parent-hash mask was a byte-pattern match over the whole RLP tree, and drift failures gave no indication of cause.

• Parse both fixture trees with BlockchainEngineXFixture/BlockchainEngineFixture; a fixture that fails typed validation is a loud error naming file and test id, never a silent degradation of the check.
• Decode BALs with BlockAccessList.from_rlp and mask only the EIP-2935 history-contract write whose value equals the payload's own parent hash (integer comparison, so the leading-zero trimming special case disappears). A parent-hash-valued write to any other account is drift again, and there is a regression test for it.
• Classify and aggregate drifts by cause: a leaked account is traced through the sibling fixture's pre and the pre_alloc group file to the address, the group, and the tests that packed it; identical causes collapse into one diagnosis instead of one failure per fixture.
• verify_engine_x_execution always returns a result with an explicit skip_reason, simplifying the pytest_sessionfinish caller; the not-clean-exit gate now logs that the check was skipped.
• The unit tests build real serialized fixtures via the models instead of hand-rolled payload dicts, and guard the field names the check depends on so model renames fail in CI.

Example failure output after injecting a leak into a real fill:

1 of 13 Engine X fixtures execute differently against their packed pre-allocation group's genesis than against their own pre-allocation (1 distinct cause(s)):

[1x] payload 0: account 0xf6c3...b3ff appears in the packed fixture's BAL only
  tests: tests/amsterdam/eip7928_block_level_access_lists/test_block_access_lists_eip2935.py::test_bal_2935_empty_block[fork_Amsterdam-blockchain_test_engine_x]
  the account is absent from the test's own pre-allocation but present in pre-alloc group 0xa58f4980b4cd5258 (11 tests, e.g. ...): an account introduced by packing leaked into this test's execution. Isolate the test with @pytest.mark.pre_alloc_group("separate") or declare the account in its pre-allocation

Validated with tests/amsterdam/eip7928_block_level_access_lists filled at Amsterdam with --generate-all-formats: 1013 Engine X fixtures execute identically against their packed group's genesis, including the corrupted-BAL negative tests; injected gas and BAL-leak drifts produce the diagnostics above and the output verifies green again once restored.

Related Issues or PRs

Stacked on #3219 (includes its commits until it merges; only the last commit is new). Addresses marioevz's engine_x_checks.py review comments on #3219.

Checklist

• Ran fast static checks to avoid CI fails, see Code Standards & Verifying Changes: just static
• PR title has the form <type>(<area>): <title>, where <type> and <area> come from an appropriate C-<type>, respectively A-<area>, label. The title should match the target squash commit message.

Cute Animal Picture

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 93.42%. Comparing base (946f45c) to head (a411c77).
⚠️ Report is 2 commits behind head on forks/amsterdam.

Additional details and impacted files

@@               Coverage Diff                @@
##           forks/amsterdam    #3265   +/-   ##
================================================
  Coverage            93.42%   93.42%           
================================================
  Files                  625      625           
  Lines                37032    37032           
  Branches              3385     3385           
================================================
  Hits                 34597    34597           
  Misses                1666     1666           
  Partials               769      769           

Flag	Coverage Δ
unittests	93.42% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.