feat(test-fill,test-specs): add t8n call caching to test specs

[danceratopz]

🗒️ Description

Add an in-memory LRU cache for transition tool outputs during fixture generation. When multiple fixture formats share the same t8n inputs (e.g., blockchain_test and blockchain_test_engine), the cache eliminates redundant t8n calls.

Changes

• Add T8nOutputCache class with bounded LRU eviction.
• Add strip_fixture_format_from_nodeid() to derive consistent cache keys.
• Integrate cache lookup/store in BlockchainTest._generate_block_data().
• Add xdist_group markers to keep related formats on the same worker.
• Sort test items during collection for deterministic execution order.

Performance

Command (with xdist), note this also generates the state_test format, which doesn't benefit from caching:

hyperfine --warmup=1 --runs=4 "taskset -c 0,2,4,5,6,7 uv run fill tests/shanghai --output=fixtures-with-cache-without-xidst --clean -x --no-html --skip-index"

Benchmarked on tests/shanghai (1408 tests across all forks through Osaka) with 6 parallel workers:

Configuration	Time	Improvement
Sequential, no cache	61.7s	baseline
Sequential, with cache	47.6s	1.3x faster
Parallel, no cache	38.7s	1.6x faster
Parallel, with cache	33.2s	1.9x faster

Sequential:

No cache   ████████████████████████████████████████  61.7s
+ Cache    ███████████████████████████████·········  47.6s  -23%

Parallel (6 workers):

No cache   █████████████████████████···············  38.7s  -37%
+ Cache    ██████████████████████··················  33.2s  -46%

• Percentages relative to sequential baseline (61.7s).
• So using xdist, the cache gives a ~1.17x speed-up or a ~14.2% improvement.

🔗 Related Issues or PRs

N/A.

✅ 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).
• Tests: Ran mkdocs serve locally and verified the auto-generated docs for new tests in the Test Case Reference are correctly formatted.

Cute Animal Picture

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 86.07%. Comparing base (7f584cd) to head (f57d1d5).
⚠️ Report is 3 commits behind head on forks/amsterdam.

Additional details and impacted files

@@               Coverage Diff                @@
##           forks/amsterdam    #2084   +/-   ##
================================================
  Coverage            86.07%   86.07%           
================================================
  Files                  599      599           
  Lines                39472    39472           
  Branches              3780     3780           
================================================
  Hits                 33977    33977           
  Misses                4862     4862           
  Partials               633      633           

Flag	Coverage Δ
unittests	86.07% <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.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[marioevz]

Regarding this comment, we are currently counting opcodes every single time, but we could improve this by doing so only for tests that are marked to require it.

We could mark the tests that need this information with a new marker (pytest.mark.count_opcodes or similar). Perhaps we can set it at a folder level, e.g. for all tests in tests/benchmarking.

    if request.node.get_closest_marker("count_opcodes"):
      session_t8n.reset_opcode_count()
    else:
      session_t8n.remove_opcode_count()

or similar.

[marioevz]

Created tracking issue so we can merge without this: #2102

[marioevz]

Running the latest version I see that we are adding @t8n-cache-<md5-hash> to the names of all tests for some reason. This seems incorrect to me, so I'd like to give this a proper re-review.

[danceratopz]

Running the latest version I see that we are adding @t8n-cache-<md5-hash> to the names of all tests for some reason. This seems incorrect to me, so I'd like to give this a proper re-review.

This is to ensure that all parametrized test formats (state_test, blockchain_test, blockchain_test_engine) for a single test case get distributed to the same xdist worker to ensure they use the same per-worker cache; the cache is not global across workers.

[SamWilsn]

Just a general comment (haven't looked at the code yet), but are the caches per-thread/per-process or global? If they aren't global, you might need to load group them to get the biggest benefit.