feat(tests, spec-specs): add full cost per state byte pricing function for EIP-8037

[spencer-tb]

🗒️ Description

Summary

• Un-hardcode the cost_per_state_byte pricing function in both the spec and testing framework to compute the EIP-8037 quantized value from the block gas limit
• Add BaseFork._env_gas_limit class variable, synced automatically by the fork fixture from Environment.gas_limit (default 120M, or --block-gas-limit)
• All downstream gas methods (gas_costs(), sstore_state_gas(), etc.) respond to the actual block gas limit — no parameter threading needed
• Tests that set Environment(gas_limit=X) or use --block-gas-limit get the corresponding cost_per_state_byte automatically
• Add parametrized tests across 10 block gas limits (1M to 1G) for SSTORE, CREATE, CALL, SELFDESTRUCT, auth, and refunds
• Add underflow boundary tests at low gas limits where the quantized result falls below the offset, verifying the floor of 1

Gas Limit Flow

1. Session start, filler plugin reads EnvironmentDefaults.gas_limit (default 120M, or --block-gas-limit) via the env_gas_limit fixture
2. fork fixture, creates a per-test fork variant via fork.with_env_gas_limit(env_gas_limit) and monkey-patches Environment so any Environment(gas_limit=X) syncs X back to the variant's _env_gas_limit
3. Test code, calls fork.gas_costs(), fork.sstore_state_gas(), etc. which read cls._env_gas_limit to compute cpsb
4. Fill pipeline, state.py / blockchain.py set fork._env_gas_limit = env.gas_limit per-block before t8n execution
5. EELS spec, state_gas_per_byte(gas_limit) receives the block gas limit from the environment and computes cpsb dynamically

Fork Calculator Split

On top of this, Mario's refactor splits the previously combined opcode-gas surface on BaseFork into three separate per-dimension maps:

• opcode_state_map — state gas cost per opcode (cpsb multiplier or callable)
• opcode_refund_map — regular gas refund per opcode
• opcode_state_refund_map — state gas refund per opcode

Tests can now query each dimension independently instead of pulling a mixed regular+state value. Paired with the dynamic cpsb, opcode state-gas/refund calculations scale with the block gas limit automatically.

Remaining Fixes

A separate follow-up commit (feat(tests): remaining fixes) addresses pre-existing tests that assumed the previously-hardcoded cpsb=1174 and broke under the new dynamic computation:

• test_state_gas_reservoir.py: pin fork._env_gas_limit in four tests so every cpsb-dependent value read matches what the block uses at execution time
• test_bls12_variable_length_input_contracts.py: apply the default env gas limit variant in the discount-table splitter so collection-time cpsb matches runtime

Ported Static Test Fixes

A separate commit (fix(tests): drop inflated env gas_limit in ported_static) removes the gas_limit=<value> line from Environment(...) calls in 576 ported_static test files. These values were inherited as boilerplate from the original static JSON fillers (ranging from 1M to max-i64) with no functional purpose, under the old hardcoded cpsb=1174 they didn't matter, but under dynamic cpsb they caused:

• Inflated values (1B-100B, max-i64): cpsb ballooned to tens of thousands, making SSTORE 0 to nonzero OOG with the standard tx budget
• Deflated values (1M-30M): cpsb dropped to 1–150, shifting gas consumption enough to break Storage / Account.balance assertions

Removing the override lets Environment default to EnvironmentDefaults.gas_limit = 120M, which yields the canonical cpsb=1174 these tests were always implicitly designed against. All 576 files verified to have a single Transaction with tx.gas_limit <= 14M and no references to env.gas_limit in test logic, so the change preserves test semantics.

🔗 Related Issues or PRs

N/A.

✅ Checklist

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

  just static

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

[kclowes]

LGTM!

[marioevz]

Just two preliminary comments, which I think we could even apply directly to forks/amsterdam.

[marioevz]

@spencer-tb created #2690 to address the comment I made, it's not a pretty PR but it does the job, and I did an ad-hoc (not in the PR) verification in the specs/state.py file by comparing env.gas_limit to fork._env_gas_limit and it was exactly the same for all tests. PTAL.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
⚠️ Please upload report for BASE (eips/amsterdam/eip-8037@f5da8ef). Learn more about missing BASE report.

Additional details and impacted files

@@                    Coverage Diff                     @@
##             eips/amsterdam/eip-8037    #2687   +/-   ##
==========================================================
  Coverage                           ?   86.28%           
==========================================================
  Files                              ?      599           
  Lines                              ?    37164           
  Branches                           ?     3804           
==========================================================
  Hits                               ?    32066           
  Misses                             ?     4530           
  Partials                           ?      568           

Flag	Coverage Δ
unittests	86.28% <100.00%> (?)

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

☔ View full report in Codecov by Sentry.
📢 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.

[spencer-tb]

@marioevz

Added remaining 8037 test fixes here, with TODOs, feel free to check over: 62719d8

Ported static fixes here (updated in PR desc): a3cbac3

PR desc includes the fork calculator split. We can consolidate in future PR.

Will merge once CI finally passes...