…se (ethereum#2363)

* feat(spec-specs): update EIP-8037 to match latest spec revision

* feat(tests): add EIP-8037 state creation gas cost increase tests

Co-authored-by: Sam Wilson <57262657+SamWilsn@users.noreply.github.com>

…e pattern

Co-authored-by: marioevz <11726710+marioevz@users.noreply.github.com>

Co-authored-by: marioevz <11726710+marioevz@users.noreply.github.com>

Co-authored-by: kclowes <6540608+kclowes@users.noreply.github.com>

…-by-design tests (ethereum#2843)

* feat(forks): add Fork.oog_budget_lift helper for EIP-8037 OoG tests

OoG-by-design ported_static tests are calibrated to land just below
a cost boundary; on Amsterdam each fresh SSTORE-set and CREATE
spills its state-gas portion back into regular gas when the per-tx
reservoir is empty. Tests that expect "N SSTOREs and M CREATEs
complete before OoG" need their gas budget lifted by
N * sstore_state_gas + M * create_state_gas to land at the same
intermediate state.

`Fork.oog_budget_lift(sstores_before_oog=N, creates_before_oog=M)`
composes the two existing state-gas helpers and returns 0 on
pre-EIP-8037 forks (where both helpers are 0), so callers can apply
it unconditionally without a fork guard.

Unit-tested on Cancun (zero) and Amsterdam (cumulative spill).

* fix(ported_static): Approach-1 fork-conditional OoG lift for stCreate*/stRevert*/stWallet* tests

Eight tests share the OoG-by-design shape `tx_gas = [oog_path, success_path]`
where the success_path budget is tuned to barely complete a single
CREATE, a CREATE plus a few SSTOREs, or a deploy chain. On Amsterdam
EIP-8037 splits NEW_ACCOUNT, fresh SSTORE-set, and code-deposit cost
into a regular portion plus a state-gas portion; with an empty
reservoir, the state-gas spills back into regular gas and breaks the
success_path budget.

Apply `Fork.oog_budget_lift` with the right (creates, sstores,
deploy_code_size) counts to lift the budget on Amsterdam only. Pre-
EIP-8037 forks return 0 from the helper, so the original budget is
preserved.

Files (skip-list entries cleared):
- stCreate2/test_create2_oo_gafter_init_code.py (-g1)
- stCreate2/test_create2_oo_gafter_init_code_returndata2.py (-g1)
- stCreateTest/test_create_oo_gafter_init_code.py (-g1)
- stCreateTest/test_create_oo_gafter_init_code_returndata2.py (-g1)
- stRevertTest/test_revert_sub_call_storage_oog.py (-g1-v0)
- stRevertTest/test_revert_sub_call_storage_oog2.py (-g1-v0)
- stWalletTest/test_wallet_construction_oog.py (-g1)
- stWalletTest/test_multi_owned_construction_not_enough_gas_partial.py (-g1)

Removes 8 entries from amsterdam_skip_list.txt.

* chore(ported_static): remove stale Amsterdam skip entries for stSStoreTest 16-pair family

22 stSStoreTest files (`sstore_0to*`, `sstore_xto*` excluding `gas`,
`gas_left`, `change_from_external_call_in_init_code`) had 3 skip
entries each (`d{0,1,2}-g1`). Re-running these on Amsterdam with the
skip list disabled shows all 60 fixture variants per file already
pass without any code changes.

The post-state expectation at `g=1` is `contract_2: storage={1: 0}`
(only slot 1, asserted to be zero). On Cancun, ~14 of the 16
SSTORE pairs complete before OoG; on Amsterdam EIP-8037 the state-
gas spill cuts that to ~5 pairs. In both cases slot 1 ends at 0 and
the final `SSTORE(1, 1)` does not run, so the assertion holds on
both forks unchanged.

These were defensive skip entries from an earlier snapshot. Remove
all 66 entries; no test-file changes needed.

* fix(ported_static): lift tx_gas[0] on Amsterdam for the 2 base oo_gafter_init_code tests

Address kclowes's review on ethereum#2843: with only tx_gas[1] lifted, g=0
OoG'd at CREATE/CREATE2 dispatch on Amsterdam (NEW_ACCOUNT state-gas
spill) before init code ever ran — the assertion still held
(`NONEXISTENT` either way) but the failure mode shifted from "OoG
after init code" (the test's named scenario) to "dispatch-time OoG".

A clean closed form using `fork.oog_budget_lift(creates_before_oog=1)`
(183600) overshoots and pushes g=0 past the deploy threshold. The
Cancun 1000-gas gap between g=0 and g=1 collapses on Amsterdam:
once dispatch is cleared, the 5-byte init code is cheap enough to
always complete. Empirical binary-search on both files puts the safe
range at (166499, 167000); 166_750 sits in the middle, keeping g=0
OoG'ing at dispatch and g=1 just clearing the deploy threshold.

The two `_returndata2` variants are left unchanged — g=0's post-state
happens to land identically on both forks at the existing budget, and
adding any lift breaks them.

* fix(spec-specs, test-types): resolve lint errors from forks/amsterdam → devnets/bal/7 merge

The May-18 merge `dffc4cfea` ("Merge remote-tracking branch
'upstream/forks/amsterdam' into devnets/bal/7") had two conflict
resolutions that left `bal/7` in a state where `just static` fails:

1. `src/ethereum/forks/amsterdam/blocks.py` ended up with the EIP-7843
   `slot_number: U64` field declared twice (lines 260 and 268).
   `mypy` rejects it with `[no-redef]`; `ethereum-spec-lint` crashes
   with `ValueError: duplicate path Header.slot_number`. Remove the
   second copy.

2. `BuiltBlock.derive_engine_payload_modifier` was dropped from
   `packages/testing/src/execution_testing/specs/blockchain.py` while
   its 5 call sites in `specs/tests/test_types.py` were kept. `mypy`
   reports 5 `attr-defined` errors. Restore the staticmethod (and the
   `FixtureExecutionPayloadModifier` import it needs) from
   `forks/amsterdam`.

The instance-level wiring on `forks/amsterdam` (`BuiltBlock.rlp_modifier`
field, constructor pass-through, and `get_fixture_engine_new_payload`
call) is **not** restored — neither the linter nor the test file
references it, and `bal/7`'s current `get_fixture_engine_new_payload`
already runs without it.

This unblocks CI on every open PR against `devnets/bal/7` (including
this one).