test(eth): cover accessList formatter on type 1, 2 and 4 transactions

[genisis0x]

What

Closes #3707.

Issue #3707 flags that the transaction-param formatters lack direct tests for several accessList combinations the wire format actually allows. Add three focused cases (with async + sync mirrors) that exercise the eth_sendTransaction and eth_sendRawTransaction paths together.

New test	Tx type	Path	Pins
test_(async_)eth_send_raw_transaction_type_1_access_list	EIP-2930 (type 1)	eth_sendRawTransaction	gasPrice + accessList, no max-fee fields; the round-tripped tx reports type == 1 and preserves the access-list entries
test_(async_)eth_send_transaction_type_2_access_list	EIP-1559 (type 2)	eth_sendTransaction	maxFeePerGas + maxPriorityFeePerGas + accessList go through separate formatter branches; the combination is the regression case
test_(async_)set_code_transaction_with_access_list	EIP-7702 (type 4)	eth_sendRawTransaction	authorizationList + accessList survive together on the raw signing path; the existing set-code test only exercises authorizationList

Why these three

The existing coverage at web3/_utils/module_testing/eth_module.py:

• test_eth_create_access_list exercises the eth_createAccessList RPC and then feeds the result into eth_sendTransaction (type 2 by default), but doesn't drive eth_sendRawTransaction and doesn't lock the tx type explicitly.
• test_(async_)sign_authorization_send_raw_and_send_set_code_transactions exercises authorizationList on both raw + send paths, but never with accessList attached — so a regression in the type-4 formatter that drops accessList would not be caught.

The three additions close exactly the gaps the issue points at without changing production code.

No production changes

Pure test additions:

• web3/_utils/module_testing/eth_module.py — 6 new methods (3 async, 3 sync mirrors).
• web3/_utils/module_testing/eth_module.py — import AccessList + AccessListEntry from web3.types for typed literals (satisfies TxParams.accessList declared as AccessList, and resolves a Sequence[str] inference issue mypy raised on the dict-literal form).
• newsfragments/3707.internal.rst.

Tests

$ python -m pytest tests/integration/test_ethereum_tester.py::TestEthereumTesterEthModule \
       -k "access_list or set_code" -v
test_eth_send_transaction_type_2_access_list[<lambda>] PASSED
test_eth_send_transaction_type_2_access_list[identity] PASSED
test_eth_send_raw_transaction_type_1_access_list      PASSED
test_set_code_transaction_with_access_list            PASSED
test_eth_create_access_list[<lambda>]                 PASSED  (existing)
test_eth_create_access_list[identity]                 PASSED  (existing)
test_sign_authorization_send_raw_and_send_set_code_transactions PASSED  (existing)
7 passed

Async mirrors are not exercised against EthereumTesterProvider (no async tester runner exists for the eth module today — the async eth-module suite runs against the geth integration), but the implementations are direct ports of the passing sync cases with await added; the geth CI pipeline will exercise them.

$ pre-commit run --files web3/_utils/module_testing/eth_module.py
all hooks green (black, flake8, isort, pydocstyle, mypy, autoflake, pyupgrade, blocklint)

Open question for the maintainers

Issue #3707 also wonders aloud about blob (type 3) transaction coverage. The current eth-tester backends don't appear to validate type-3 round trips end-to-end the way they do for the other types, so I've left that out of scope here. Happy to add it as a follow-up if a tester path is available, or punt to the geth integration suite.