docs: move the process pages out of the docs top level (#901)

[JarryShaw]

• Searched for similar pull requests
• Followed the coding style (make pylint, make mypy, make isort)
• make test passes, and a test case covers the change
• Added a changelog entry under docs/source/changelog/ and regenerated CHANGELOG.md, if the change is user-visible

What is the purpose of your pull request?

• fix
• feat
• perf
• refactor
• test
• docs
• ci
• chore

---

Description of your pull request and other information

Closes #901. Moves conventions.rst, releasing.rst, testing.rst, workflows.rst and pep.rst into a new docs/source/contributing/ subdirectory via git mv, one commit. demo.rst, ext.rst and changelog.rst stay at the top level (library-usage docs, not process); index.rst is the root document.

Fixed every reference the move touched: index.rst's toctree, the one cross-tree :doc: link that broke (pcapkit/protocols/internet/mh.rst → /pep), pep.rst's own eleven :doc: links, which gained a leading / -- ten into pcapkit/, plus one to /ext, a page that stays at the root, and the plain-text path mentions in CONTRIBUTING.md, README.md, a workflow comment, and four test docstrings/comments.

Splitting the toctree moves changelog from last to third in the rendered nav.

Sphinx build: 58 warnings, same as the 58-warning baseline (delta 0). util/changelog_md.py --check exits 0 (unaffected -- changelog.rst didn't move). The real setuptools sdist ships the same 928 entries before and after, with identical tar tzf listings. Six of those files do change content -- README.md, both PKG-INFO copies and the three tests/ modules, since MANIFEST.in's prune test does not match tests/ -- all of it comment, docstring and URL text. make isort clean; make pylint/make mypy carry their pre-existing, advisory findings, none in a file this PR touches (no pcapkit/*.py changed). Ran tests/project/ (178 tests) plus the touched unit tests rather than the full suite, which OOMs locally.

Changelog: N/A -- changelog centralised in #657.

[JarryShaw]

Good to merge — cross-review on a different model (sonnet; authored by opus) came back needs changes, prose only, and I have made exactly those fixes. The tree is byte-identical to the reviewed one (git diff c38640aa4 c23963171 is empty); only the commit message and PR body changed.

What it verified: four of the five moves are R100 byte-identical blobs, pep.rst is R098 for its reference edits alone, and a repo-wide sweep for :doc:/:file:/:ref:/bare-path/toctree/rendered-.html forms found zero broken or misdirected references. Sphinx warning delta 0 (58 → 58, from Sphinx's own summary lines, with the two warning sets identical line-for-line after normalising the tree prefix). The six relative sibling refs in releasing.rst and workflows.rst resolve in the built HTML, which is the mechanical reason pep.rst and workflows.rst had to move with the owner's three rather than a matter of taste.

What it disputed, all three confirmed by me directly and now corrected:

• The message claimed pep.rst's refs to ext/releasing/workflows "needed no change, since those move together". ext.rst does not move, its reference did change (:doc:`ext``` → :doc:/ext```), and releasing/workflowsare not referenced frompep.rstat all — 0 hits. Eleven refs changed, not ten, and six of the eleven are outsideconst/`.
• "Nothing shipped changed" proved only that the sdist file list is unchanged (928 entries, identical tar tzf). Six shipped files change content, because MANIFEST.in's prune test is singular and does not match tests/. Comment/docstring/URL text only.
• The toctree split moves changelog from last to third in the nav; now stated.

Unverified and stated as such: the author's intermediate 59-warning build was not reproduced, and make pylint/mypy were not re-run (no pcapkit/*.py changed).