chore: sync with upstream pingdotgg/t3code main

[yordis]

• Brings in upstream's per-signal OTEL_*_EXPORTER support, which restores the one exporter feature the fork dropped when it retired its own OTEL reader.
• The SQLite setup keeps the fork's shared migration ledger checks alongside upstream's new WAL size limit, since the ledger checks must still run before any migration.
• The routes layer keeps the fork's MCP toolkit dependency injection and places upstream's untraced requests layer last, as upstream requires, so no route layer can replace it.

---

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

[cursor]

PR Summary

Medium Risk
Touches auth replay storage, cloud link side effects, and projection/shell query paths used by background reactors; most changes are operational or client UX with targeted server tests.

Overview
This sync pulls upstream observability and triage work: t3 trace summary reads rotated server.trace.ndjson with per-span rates and percentiles; trace diagnostics stream files instead of loading them whole; an event-loop stall monitor and SIGUSR2 heap snapshots land in traces/diagnostics; DPoP/cloud replay markers are pruned on a schedule; and triage/playbook no longer assume a server.log—they point at boot-service.log and server-child*.log* plus the trace file. The OTLP trace proxy is excluded from server tracing to avoid feedback loops.

Server orchestration changes reduce background load: PR sync uses listThreadsWithPullRequests instead of hydrating the full shell; settlement/PR sweeps can read unsettledOnly shell slices via readSweepSnapshot; shell snapshots skip a redundant schema decode on the hot path. Provider session start now forwards a title only when the thread was manually renamed (not auto-title/prompt seed). SQLite adds a WAL size cap with checkpoint behavior covered by tests.

Mobile fixes include connection field accessibility (label hidden from the tree, single aria-label on inputs), composer width after Android fold resume, thread feed remount when messages arrive past a thinking placeholder, shared shell snapshot cache encoding, usage Cursor keychain detection/messaging, and thread-list PR row display tweaks. Desktop passes through OTEL_*_EXPORTER env vars to the backend child. node-pty is bumped to a beta 1.2.x release.

^{Reviewed by Cursor Bugbot for commit 4f30694. Bugbot is set up for automated code reviews on this repo. Configure here.}

[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	13.5 KiB	13.5 KiB	−10 B (−0.1%)	15.1 KiB	✅
Codex	Thread snapshot wire	7.1 KiB	7.1 KiB	+2 B (+0.0%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.4 KiB	6.4 KiB	−12 B (−0.2%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	56.2 KiB	56.2 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	9	9	0 (0.0%)	21	✅
Claude	Total thread wire	13.5 KiB	13.5 KiB	−1 B (−0.0%)	15.1 KiB	✅
Claude	Thread snapshot wire	7.1 KiB	7.1 KiB	−2 B (−0.0%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+1 B (+0.0%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	57.0 KiB	57.0 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	9	9	0 (0.0%)	21	✅

Baseline: 6fb9f79 · PR result: 4f30694 · Source CI: failure

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: 114.0 KiB
• Claude decoded thread snapshot: 114.7 KiB

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