[Mannan-Goyal]

Description

Before this change, TransactionTest.generate discarded the transition tool (del t8n). A transaction fixture therefore recorded tx.error without any check against EELS. Block-based fills can't catch this either: they send the tool the transaction as JSON, and the t8n's JSON loader (TransactionLoad) rebuilds the transaction from its fields, ignoring rlpOverride.

With this PR, when filling with the in-process EELS t8n, every transaction fixture's exact bytes are decoded by the fork's EELS code:

• Bytes EELS can't decode, including empty or truncated bytes, must have an expected error.
• Once a fork defines FrameTransaction (EIP-8141), decoded frame txs also go through validate_frame_transaction, so a frame tx it accepts must have no expected error. On forks without it, this step does nothing.
• Known limit: validate_frame_transaction applies every transaction-only rule except the chain ID, which EELS checks against the block's chain ID when it processes the transaction. A frame transaction_test whose only fault is its chain ID would therefore be reported as an unexpected success. None of the EIP-8141 transaction_test cases does this today: test_wrong_chain_id is a state_test, and the raw-RLP chain ID of 2**64 already fails decoding.

Only the accept/reject verdict is compared, not the exception class. Other t8n tools are unaffected.

The frame step is exercised for real on the EIP-8141 branch (#3878). No fork on this branch ships FrameTransaction, so the success-direction error is unit-tested with a fake fork whose decoder returns a stand-in frame transaction (test_reference_check_rejects_accepted_frame_transaction).

Documented under docs/writing_tests/types_of_tests.md → Transaction Tests.

Verified:

• every transaction_test directory fills clean --until Bogota (11265 passed);
• on the EIP-8141 branch, all 162 EIP-8141 transaction_test cases fill clean, and an EELS FrameTransaction.sender width mutation is caught by sender_19_bytes, sender_21_bytes and sender_empty;
• breaking the legacy to decoding in EELS is caught by test_valid_reencoded_transaction;
• just test-tests: 4013 passed, 0 failed; the new unit test fails (DID NOT RAISE UnexpectedExecutionSuccessError) when the success-direction raise is removed.

Related Issues or PRs

• Part of EIP-8141 Implementation Tracker: Frame Transaction #2829 (EIP-8141 implementation tracker).
• feat(tests): add missing EIP-8141 frame transaction coverage #3917 (EIP-8141 missing tests) benefits most: 59 raw-RLP cases, the nonce/frame-gas-sum bounds and the other *_transaction cases there currently fill unchanged whatever EELS does.
• The check also applies once feat(spec-specs,test-forks,tests): implement EIP-8141 frame transactions #3878 lands FrameTransaction on forks/bogota.

Checklist

• Ran fast static checks to avoid CI fails, see Code Standards & Verifying Changes: just static
• PR title has the form <type>(<area>): <description>, 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