fix(mailbox): make collaborating threads discoverable

[pseudoseed]

The mailbox picker was empty until the user typed, used literal substring matching, and silently discarded matches after the first 12. Active threads could look unavailable, with no indication that more matches existed.

Show recent active threads immediately and provide a result count plus Show more threads. Search matches words across project and thread names, normalizes whitespace/case, ranks title matches first, and retains settled-thread search. Already-linked peers stay visible above the picker, including when their local thread shell is unavailable. Web/desktop and mobile use one shared selector, with the same environment boundary and existing link/unlink commands.

Validation: 15 focused candidate/sort tests passed; web and shared-runtime typechecks passed; the configured mobile tsc --noEmit check passed; targeted lint and formatting passed. Browser verification and before/after screenshots were skipped at the maintainer’s explicit request.

Model: GPT-6 Astra. Harness: Codex.

[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	14.2 KiB	13.9 KiB	−312 B (−2.1%)	15.1 KiB	✅
Codex	Thread snapshot wire	7.3 KiB	7.3 KiB	−3 B (−0.0%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.9 KiB	6.6 KiB	−309 B (−4.3%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	59.8 KiB	57.5 KiB	−2.3 KiB (−3.9%)	66.4 KiB	✅
Codex	Live turn messages	10	8	−2 (−20.0%)	21	✅
Claude	Total thread wire	14.0 KiB	13.9 KiB	−108 B (−0.8%)	15.1 KiB	✅
Claude	Thread snapshot wire	7.3 KiB	7.3 KiB	−15 B (−0.2%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.7 KiB	6.6 KiB	−93 B (−1.4%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	59.1 KiB	58.3 KiB	−840 B (−1.4%)	66.4 KiB	✅
Claude	Live turn messages	8	8	0 (0.0%)	21	✅

Baseline: e74ca52 · PR result: db778cf · 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: 116.2 KiB
• Claude decoded thread snapshot: 116.9 KiB

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