refactor(sync): remove or justify remaining any usages in sync tests (Phase 1.9)

[h4yfans]

Closes Phase 1 §1.9 of .claude/plans/tech-debt-remediation.md.

Summary

Audit of : any / as any across apps/desktop/src/main/crypto/** and apps/desktop/src/main/sync/**. Crypto production code was already clean (0 hits). This PR removes 17+ as any uses in sync test files by introducing typed test helpers.

Changes

• dirty-recovery.test.ts — let db: any → let db: DataDb, use asClientDb() helper.
• apply-item.test.ts — queue: null as any → real SyncQueueManager (cleaner than a null stub).
• task-sync-gaps-e2e.test.ts — 14 hits of deviceX.db as any → typed asClientDb() + asSyncDb() helpers.
• item-handlers/folder-config-handler.test.ts — mockQueue as any (×2) → Pick<SyncQueueManager, 'enqueue'> + makeMockQueue() helper.
• item-handlers/settings-handler.test.ts — db: {} as any → db: {} as unknown as DrizzleDb; removed invalid deviceId field, replaced with proper emit: vi.fn().

Verification (run locally by agent)

• grep -rn ": any\|as any" apps/desktop/src/main/crypto apps/desktop/src/main/sync → 0 hits
• npx tsc -p apps/desktop/tsconfig.node.json --noEmit — no errors
• npx tsc -p apps/desktop/tsconfig.web.json --noEmit — no errors
• vitest run on affected files → 59/59 passing
• vitest run on full src/main/sync → 627/627 passing across 52 files

Test plan

• typecheck:node clean
• typecheck:web clean
• all sync tests passing
• CI lint passes

🤖 Generated with Claude Code