chore(turbo): ingest upstream main through 1a003e383 (nightly .1042), preserve all seams

[gfsaaser24]

Brings turbo current with upstream pingdotgg/t3code main tip 1a003e383 — all 21 missing commits, covering nightlies .1038 through .1042 plus the un-released main tip. Every Turbo seam verified preserved: 9/9 seams, 88 file checks.

Notable inbound

• e70cdb478 — our own fix(server): stop Claude resume handshakes from completing turns that never ran pingdotgg/t3code#5710 merged upstream (phantom resume-handshake turns). Supersedes turbo's downstream copy; the merge takes upstream's squash, which also includes the lint-ratchet fix — makes fix(server): stop the phantom-turn tests from tripping the lint ratchet #45 redundant (close it when this lands).
• 49964e38c — upstream vouched gfsaaser24.
• ba9c9ae81 — one greedy agent process no longer takes down the whole server.
• 05eb05118 / 288d8e345 / 076e9048d — draft-preservation sidebar rework, project settings page, manual project icons.
• 6dbffa022 — worktree vs current-checkout per project.

Conflicts resolved (5 files)

File	Resolution
ProviderRuntimeIngestion.test.ts	Took upstream verbatim (upstream squash includes our lint fix; verified byte-identical to upstream)
settings.ts (contracts)	Keep-both import (ScopedThreadRef + ThreadEnvMode)
Sidebar.tsx	Keep-both imports (upstream draft exports + pane seam imports)
useHandleNewThread.ts	Real seam rework: upstream pingdotgg#5777's invested-draft-reuse semantics threaded through the pane seam's sourceTarget/navigation/DraftId contract; useNewThreadHandler compat wrapper untouched so upstream call sites work unchanged
routeTree.gen.ts	Keep-both routes (/settings/t3-turbo + /settings/projects/$projectKey)

Also in this PR (unblocks the nightly pipeline)

• Seam remap: upstream 0de954073 renamed SidebarV2.tsx → Sidebar.tsx; the pane marker verifiably lives on at Sidebar.tsx (line ~2930). Dropped the dead duplicate manifest entry + updated its test. This was failing all manifest verification since the rename (red Test on fix(server): stop the phantom-turn tests from tripping the lint ratchet #45, and it would have failed every nightly candidate).
• Re-anchored .t3-turbo/upstream.json: it recorded the 08-07 anchor while turbo actually contained .1035, and its version no longer derived from its nightlyTag — the sync validator rejects that, matching the 08-08 nightly-sync failure. Now anchored to exactly what this merge ingests.

Verification

• Turbo customization manifest: 9 seams, 88 checks pass
• vp check: 0 errors
• typecheck (@t3tools/web, @t3tools/contracts): clean
• Chat-pane seam + hooks tests: 60/60
• Manifest + nightly-sync script tests: 28/28
• ProviderRuntimeIngestion 47/47 + ProviderSessionReaper 9/9 (one combined-invocation run flaked on machine-load timeouts, clean on rerun; both suites individually green)

No version bumps in this PR — versioning stays on our own line per policy.

Model and harness: Claude Opus 5 (1M context) in Claude Code.

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

ℹ️ No successful main baseline artifact is available yet. This run establishes the initial measurement.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	—	11.3 KiB	—	15.1 KiB	✅
Codex	Thread snapshot wire	—	5.5 KiB	—	7.3 KiB	✅
Codex	Live turn WebSocket wire	—	5.9 KiB	—	7.8 KiB	✅
Codex	Live turn WebSocket decoded	—	49.7 KiB	—	66.4 KiB	✅
Codex	Live turn messages	—	16	—	21	✅
Claude	Total thread wire	—	11.3 KiB	—	15.1 KiB	✅
Claude	Thread snapshot wire	—	5.5 KiB	—	7.3 KiB	✅
Claude	Live turn WebSocket wire	—	5.8 KiB	—	7.8 KiB	✅
Claude	Live turn WebSocket decoded	—	50.6 KiB	—	66.4 KiB	✅
Claude	Live turn messages	—	16	—	21	✅

Baseline: unavailable · PR result: da968fc · 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: 94.6 KiB
• Claude decoded thread snapshot: 95.4 KiB

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