feat(ci): split release builds across multiple runners; delete develop; stable becomes mainnet

[danceratopz]

🗒️ Description

Split multi-fork fixture releases across parallel runners by fork range, then simplify the release strategy by collapsing the two release workflows into one.

Summary of changes

• Multi-fork releases (e.g., mainnet) are split across parallel self-hosted runners by fork range (pre-cancun, cancun, prague, osaka, ...), then combined into a single tarball.
• The develop feature is removed; stable is renamed to mainnet.
• The full release workflow (release_fixture_full.yaml) is deleted. All releases now use release_fixture_feature.yaml with tests-<feature>@v* tags.
• Mainnet releases are marked as regular releases (not pre-releases). Other features (bal, benchmark) remain pre-releases.

How we got here

The initial goal was to reduce release CI wall-clock time by splitting expensive fill jobs across parallel runners, following the pattern from #2529.

While implementing the split, we noticed that stable and develop would share identical fork-range builds (same EELS implementation, same fill-params minus --until). The deduplication logic to handle this was fragile: it silently assumed both evm-types produce the same output, and would break if they ever diverged.

This led to a simpler realization: the develop release is redundant. The bal feature already fills Amsterdam for devnets, and the full release workflow exists only to bundle stable + develop. Removing develop eliminates the cross-feature deduplication problem entirely and lets us collapse the full and feature release workflows into one. A mainnet release is just another feature release (tests-mainnet@v6.0.0).

Workflow structure

Before:

features -> build (matrix: feature) -> release

After:

setup -> build (matrix: fork-range) -> combine -> release

The setup job reads feature.yaml and emits a flat build matrix of fork ranges. Each runner fills its range to a directory. The combine job merges the directories and creates the final tarball. Features without fork ranges (e.g., benchmark) skip the combine step and produce the tarball directly.

Config changes

• feature.yaml: top-level fork-ranges list defines how multi-fork features are split. Features with --until in their fill-params are automatically split across applicable ranges.
• evm.yaml: stable/develop entries replaced with single eels entry.
• New feature_only: true flag excludes features (benchmark, bal) from --all mode (now removed, but the flag remains useful as documentation).

New scripts

• generate_build_matrix.py: reads feature.yaml, emits build matrix and combine metadata for GitHub Actions.
• create_release_tarball.py: creates a release tarball from a merged fixture directory, using pigz for parallel compression when available.
• Both scripts have integration tests in .github/scripts/tests/ (runnable via just test-ci-scripts).

PRs and Actions

• chore(ci): enable ported static in py3 and split into 5 parallel fork-range runners #2529: same pattern applied to py3 tox env.
• First successful split build (stable + develop): validates the fork-range splitting approach before the simplification to mainnet-only.
• First mainnet build WIP.

✅ Checklist

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

  uvx tox -e 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).

Cute Animal Picture

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 86.24%. Comparing base (1e49423) to head (56b3910).
⚠️ Report is 2 commits behind head on forks/amsterdam.

Additional details and impacted files

@@               Coverage Diff                @@
##           forks/amsterdam    #2592   +/-   ##
================================================
  Coverage            86.24%   86.24%           
================================================
  Files                  599      599           
  Lines                36984    36984           
  Branches              3795     3795           
================================================
  Hits                 31895    31895           
  Misses                4525     4525           
  Partials               564      564           

Flag	Coverage Δ
unittests	86.24% <ø> (ø)

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.

[danceratopz]

Ah @spencer-tb, this is probably missing an index.json!!!