docs: second-round pass over the remaining Sphinx pages and root docs (#719)

[JarryShaw]

• Searched for similar pull requests
• Followed the coding style (make pylint, make mypy, make isort) — N/A, docs-only (Sphinx -n and tests/project below)
• make test passes, and a test case covers the change — tests/project passes unedited
• 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

Round two of #719 over every docs/source/ page outside protocols/ and changelog/, plus README.md and CONTRIBUTING.md. Claims checked against origin/main; fixes where the code disagrees:

• pcapkit/index.rst: CLI help block regenerated from pcapkit-cli -h (adds -B/-O and stdin -); emoji is optional, not required — __main__ warns EmojiWarning and prints plain text.
• ext.rst: reassembly callbacks run per submission with that submission's datagram list; flow-tracing callbacks get one Index per finalised flow. Dumper.kind is the format name, not the file extension.
• foundation/reassembly/tcp.rst: names reassembly.tcp.TCP (not Reassembly), adds its .. module::, and the packet example matches toolkit.pcap.tcp_reassembly. ipv6.rst buffer lists timestamp.
• pep.rst: LINUX_SLL2 and QUIC also lack stubs; checksum half waits on the pseudo-header class (already settled), not an open question.
• process.rst (five pages precede it), index.rst (cd PyPCAPKit), CONTRIBUTING.md (vermin's 3.11 is detected, not the floor; release type in use).
• Timed context cut in registry-protocol, protocol-layer-placement, sentinel-convention, documentation, releasing, workflows, pep.

Sphinx -b dummy -n, fresh output dirs, these pages: 66 warnings on main, 65 here, none new (the tcp.rst py:mod one is gone).

Every page in the slice has been read. On the templated const/ and vendor/ pages the per-registry boilerplate was checked by -n, the hand-written prose by reading.

[JarryShaw]

Cross-review verdict on 86315bb66: NEEDS CHANGES (ran on Sonnet; author Opus). There is one fix to make.

foundation/reassembly/tcp.rst:164-184 adds raw_len = len(tcp.packet.payload) but never uses it. len, last and payload are still built from tcp.raw_len and tcp.raw, and neither attribute exists. toolkit/pcap.py:168-187 uses raw_len and bytearray(tcp.packet.payload), so the example should match it.

Everything else checks out:

• The CLI help block matches python -m pcapkit -h exactly.
• emoji really is optional (__main__.py:29-33, :136, :144).
• ext.rst describes the callbacks correctly: per submit() at ip.py:371 and tcp.py:551, and per finalised Index at traceflow/tcp.py:455.
• Dumper.kind is the format name, with the extension coming from ext=.
• The new .. module:: is unique.
• Buffer.timestamp exists.
• The pep.rst stub, DTLS and checksum claims match the code.
• tests/project passes 379.

Nit: emoji output appears only under -v. UNVERIFIED: the Sphinx build, which the reviewer did not run.

[JarryShaw]

Cross-review verdict on 5d184056b: GOOD TO GO (ran on Sonnet; author Opus, round 2)

The delta from 86315bb66 touches only tcp.rst and index.rst.

• The tcp.rst packet example now matches toolkit/pcap.py:162-187 field for field, and no tcp.raw_len or tcp.raw remains on the page. I re-checked this against the code myself.
• The emoji note now says it decorates only the -v output, which matches __main__.py:133-146.
• The round-1 findings stand. See the previous verdict.

UNVERIFIED: no Sphinx build. tests/project was run by the worker on this head (379 passed).