fix(consume): add ethereum/execution-specs to SUPPORTED_REPOS

[edg-l]

Summary

FixtureDownloader.get_cache_path versions the on-disk cache directory by release tag only when the release URL's owner/repo appears in SUPPORTED_REPOS. Otherwise it falls through to the unversioned cache_folder / "other" / <archive_name> path.

When the BAL fixture release was moved from ethereum/execution-spec-tests to ethereum/execution-specs for tests-bal@v7.1.0, the new URL stopped being recognized as a release URL. A v7.0.0 download from a prior session, cached at other/fixtures_bal/, kept satisfying every subsequent consume request, including ones explicitly pointing at v7.1.0. Consumers got stale fixtures with no warning.

This PR adds ethereum/execution-specs to SUPPORTED_REPOS so its release URLs hit the versioned cache path (execution-specs/tests-bal@v7.1.0/fixtures_bal/).

How this was found

While validating ethrex's bal-devnet-7 branch against tests-bal@v7.1.0:

• The downloaded tarball at https://github.com/ethereum/execution-specs/releases/download/tests-bal%40v7.1.0/fixtures_bal.tar.gz has the correct v7.1.0 BAL (sender post-balance reflects no SD-refund per feat(spec-specs, tests): remove SD state gas refunds from EIP-8037 #2845).
• The on-disk cache used by consume had the v7.0.0 BAL (still applied the SD refund).
• 76 SD-related Amsterdam tests appeared to fail with a +1,836,000 wei = +1× NEW_ACCOUNT × cpsb × gas_price mismatch; the exact amount of the removed v7.0.0 SD refund.
• The fixture file extracted from a fresh v7.1.0 tarball download matched ethrex's execution exactly. The wire-BAL the consume harness sent matched a v7.0.0 tarball's _info.url.
• Clearing ~/.cache/ethereum-execution-spec-tests/cached_downloads/ made all 2,145 amsterdam tests pass on the same ethrex commit.

So this was a stale-cache issue, not a spec or client bug. The fix prevents it from recurring once execution-specs-hosted releases become standard.

Changes

• releases.py: add ethereum/execution-specs to SUPPORTED_REPOS.
• tests/test_releases.py: add parametrized regression test for is_release_url covering every entry in SUPPORTED_REPOS, plus an explicit assertion that ethereum/execution-specs is present.

Follow-up worth considering (not in this PR)

The cache_folder / "other" / <archive_name> fallback in get_cache_path is a latent footgun: any future release hosted on a repo not in SUPPORTED_REPOS will silently collide with prior downloads sharing an archive filename. A defensive change, e.g. incorporating a hash of the URL into the "other" path, would close the door entirely. Happy to follow up in a separate PR if maintainers agree it's worth doing.

Test plan

• uv run pytest packages/testing/src/execution_testing/cli/pytest_commands/plugins/consume/tests/test_releases.py: 12/12 pass.
• uv run ruff check && uv run ruff format --check: clean.
• Manual: clear cache, re-run consume against tests-bal@v7.1.0, observe 2,145/2,145 pass on ethrex bal-devnet-7.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.01%. Comparing base (5c6cec7) to head (7875f05).

Additional details and impacted files

@@               Coverage Diff                @@
##           forks/amsterdam    #2849   +/-   ##
================================================
  Coverage            90.01%   90.01%           
================================================
  Files                  539      539           
  Lines                32618    32618           
  Branches              3030     3030           
================================================
  Hits                 29361    29361           
  Misses                2699     2699           
  Partials               558      558           

Flag	Coverage Δ
unittests	90.01% <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.

[marioevz]

LGTM, thanks for this!