feat(pull-requests): link GitHub references, add list filters, defer line stats

[rynfar]

Three upstream commits for the pull-request surface.

• #8812 link GitHub references in markdown — #123, owner/repo#123, and
  commit SHAs become links
• #8809 add pull request list filters
• #6471 defer pull request line stats until visible

#6471 is the notable one: _chat.pull-requests.tsx loses 181 lines and
pullRequestList.logic.ts loses 73, because additions/deletions are now fetched
per-row on visibility instead of eagerly for the whole list.

Carve-out

#8812 introduces a linkChildren element that already wraps children in
MarkdownLinkContext, and passes it into MarkdownExternalLinkContent. Pylon
deliberately wraps the result rather than the children, with a comment saying
why: MarkdownExternalLinkContent inspects its first child for the leading text
it splits with <wbr/>, and an element in that position makes the whole label one
unbreakable run that overflows the chat column.

Adopting upstream's shape verbatim reintroduced exactly that — a long external
link label stopped wrapping. ChatMarkdown.test.tsx's "keeps external link labels
breakable when they mix text and formatting" caught it. The autolink condition
(!isPullRequestAutolink) is adopted; the wrapping order stays Pylon's.

Verification

@t3tools/web typecheck clean, lint clean, 375 pull-request and markdown tests,
3388 web tests passing.

One pre-existing failure is unrelated and reproduces on pylon with this branch
stashed: src/terminal/ghostty/runtimeAbi.test.ts fails to parse a
.wasm?inline import locally.

Model: Claude Opus 5 in Claude Code.

---

^{Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.6 KiB	13.7 KiB	+191 B (+1.4%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+6 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.6 KiB	6.8 KiB	+185 B (+2.7%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	57.2 KiB	58.7 KiB	+1.5 KiB (+2.6%)	66.4 KiB	✅
Codex	Live turn messages	10	11	+1 (+10.0%)	21	✅
Claude	Total thread wire	13.8 KiB	13.6 KiB	−182 B (−1.3%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	−7 B (−0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.8 KiB	6.6 KiB	−175 B (−2.5%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	59.5 KiB	58.1 KiB	−1.5 KiB (−2.5%)	66.4 KiB	✅
Claude	Live turn messages	11	10	−1 (−9.1%)	21	✅

Baseline: 9378c6e · PR result: 7c11248 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 109.5 KiB
• Claude decoded thread snapshot: 110.2 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.