feat(ai-partner): continue-in-tab handoff from peek (#1464)

[CraigBuckmaster]

Summary

Resolves phase 3, card #1464 of epic #1446. Completes the FAB-peek → Amicus tab story arc.

• New promotePeekToThread service generates a word-boundary-truncated title from the first user message, creates the thread row in amicus_threads, replays each peek message into amicus_messages (citations + follow_ups serialized), and navigates via parent.navigate('AmicusTab', { screen: 'Thread' }).
• AmicusPeekSheet wires the CTA to the service, tracks handoffInProgress to prevent double-taps, and surfaces "Couldn't save conversation — try again" on DB failure. On failure the peek stays open so the user can retry — nothing is lost.
• PeekMiniConversation accepts an optional handoffError string so handoff failures render cleanly without overloading the streaming-error code set.

Title generation cases:

• < 10 chars → "New Amicus conversation" fallback
• ≤ 50 chars → raw text
• > 50 chars → truncated at a word boundary with trailing … (hard-cut if no usable boundary)

Acceptance criteria

• CTA appears at turn 3 (inherited from ai-partner: mini-conversation component (inside peek) #1463)
• Tap creates thread with generated title, writes all messages
• Navigation lands on AmicusTab/Thread with the new threadId
• Peek closes after navigation
• Underlying screen state preserved (standard peek-dismissal)
• Double-tap prevented; CTA shows "Saving…" during save
• DB failure → banner + peek stays open; messages preserved
• Unit tests: title generation (5 cases), happy path, DB failure, missing parent nav, peek-integration (CTA/error/double-tap)
• No any types; lint clean

Test plan

• npx jest src/services/amicus/__tests__/promotePeekToThread.test.ts — 9 passing
• npx jest src/components/amicus/__tests__/AmicusPeekSheet.test.tsx — 8 passing (3 new peek-integration tests)
• Full suite: 3391 passing
• Coverage thresholds met: 80.53 / 67.47 / 73.06 / 82.31 (vs required 80 / 65 / 72 / 82)
• npx tsc --noEmit clean for all Amicus files
• npx eslint clean (0 errors, 0 warnings) on changed files

https://claude.ai/code/session_01Pht3kzgdvkn81DDfL9SnFe

[github-actions]

Test Results

✅ All tests passed

	Passed	Failed	Total
Tests	✅ 3391	❌ 0	3391
Suites	✅ 461	❌ 0	461

Coverage

Statements	Branches	Functions	Lines
—	—	—	—

⏱️ Duration: 80.9s