docs(contributing): tighten the prose from the post-sweep merges (#719)

[JarryShaw]

• Searched for similar pull requests
• Followed the coding style (make pylint, make mypy, make isort) — not run; comments/docstrings only, AST-equal below
• make test passes, and a test case covers the change — tests/project (375 passed, 1 skipped) plus each touched module alone
• Added a changelog entry — N/A — centralised in docs(changelog): shared 1.5.0 changelog — long-lived, merges last (#610, #616, #617, #618, #620) #657

What is the purpose of your pull request?

• docs — documentation only

---

Description of your pull request and other information

Final #719 pass over the prose the post-#1056 merges added to docs/source/contributing/, tests/project/, util/, .github/ and pyproject.toml.

• Cuts issue/PR numbers, dates and "used to" history from new workflow, pyproject.toml and util/run_unittest_leg.py comments; rationales kept (tests/ citations left alone, being exempt).
• Fixes prose the per-class re-import made stale (coverage.toml's "thousands of tests purge in setUp"; pyproject.toml's "a generation per test").
• States the ruleset's current six required contexts (Required checks passed + Compat Python 3.10-3.14) in unit-tests.yml and workflows.rst, keeping why the aggregate and the separate Compat legs exist.
• Fixes counts: the rcfile adds three keys (parallel, patch, core), not two; test_unit_tests_queue.py lists five pins, not "three".
• Fixes unit-tests.yml's stale pyproject.toml:151 (the marker is at 157) and renumbers the unit-tests.yml:N citations in workflows.rst/releasing.rst (pinned by TestDocCitationsResolve).

Numstat: 10 files, +130 / -152.

Sphinx -n into fresh dirs, branch vs origin/main: 1293 vs 1293 warnings, none new.

File	Compared	Equal to origin/main
.github/coverage.toml	toml	True
.github/workflows/coverage-comment.yml	yaml	True
.github/workflows/project-status.yml	yaml	True
.github/workflows/unit-tests.yml	yaml	True
pyproject.toml	toml	True
tests/project/test_coverage_rcfile.py	ast	True
tests/project/test_unit_tests_queue.py	ast	True
util/run_unittest_leg.py	ast	True

[github-actions]

Coverage: 86.48% (unit tier, Python 3.14, f73b1880f, Unit Tests run success)

Package	Statements	Missed	Branches	Partial	Cover
pcapkit (top level)	104	4	20	4	93.55%
pcapkit/const	18757	1035	2342	846	90.23%
pcapkit/corekit	1874	126	578	32	92.25%
pcapkit/dumpkit	136	25	40	6	76.70%
pcapkit/foundation	2422	298	842	82	84.99%
pcapkit/interface	112	7	40	5	92.11%
pcapkit/protocols	15653	677	3942	280	94.14%
pcapkit/toolkit	487	71	144	3	84.15%
pcapkit/utilities	429	47	122	2	85.30%
pcapkit/vendor	4409	2359	1006	158	42.84%

Per-file detail: the coverage-html artifact of this run.

[JarryShaw]

Cross-review verdict on 42fd0b232: NEEDS CHANGES (one stale passage) (reviewer ran on Sonnet, author on Opus)

The reviewer found no required changes:

• The YAML, TOML and AST of every touched file parse equal to efc9c3774.
• The three rcfile keys, the timer comment, and every unit-tests.yml:N citation hold.
• The rationale is kept: per-engine jobs, step-level continue-on-error, gc, the apt cap, and queue: max.
• tests/project gives 375 passed, 1 skipped.
• Sphinx -n reports no warnings in the changed .rst files.

The reviewer also noted that the ruleset prose still describes a 22-context ruleset. I checked the live ruleset 23497679 myself: it requires 6 contexts, Required checks passed and Compat Python 3.10–3.14. So the present-tense "names 22 exact contexts" at unit-tests.yml:1234-1255 is false today. The 5+5+5+2=17 history in workflows.rst (around line 414) is timed context. Both should state the current contract instead: which jobs the aggregate covers, and why Compat stays separate.

[JarryShaw]

Cross-review verdict on f73b1880f: GOOD TO GO (round 2, ran on Sonnet; author Opus)

This round covers the ruleset-prose rewrite from 42fd0b232 to f73b1880f:

• Ruleset: the live ruleset 23497679 requires exactly the six contexts the prose names.
• needs: list: unit-tests.yml:1321 lists changes, test, integration, engine-tests, pypcap-parity, project-tests, matching the text.
• skipped: it is accepted only when the PR is docs-only (CODE=false) and project-tests succeeds.
• Compat: these checks stay separate because a job cannot needs: a job in another workflow.
• Citations: 1320, 1319-1373, 1322, 1234-1244 and 1235 all point at what they describe, and the earlier ranges are unmoved.
• Rationale kept: why an aggregate rather than one check per matrix cell, why changelog is not a gate, and why Compat is separate.
• Unchanged: YAML, TOML and AST all parse equal to efc9c3774.
• Tests: tests/project gives 375 passed, 1 skipped.
• UNVERIFIED: Sphinx was not rebuilt for this round. The delta is prose only, and round 1's -n build was clean.