feat: establish read-only Drive intelligence foundation

[miamicreme]

Purpose

Begin converting the STT executive prototype into a production-oriented, read-only information intelligence platform using the agreed spine-and-module architecture.

Implemented

• Core tenant, job, domain-event, audit-event, and telemetry contracts
• Google Drive read-only connector contract and minimum OAuth scopes
• Canonical file inventory record and repository contracts
• Paginated file inventory service
• Trashed-item filtering
• External/public permission indicators
• Failure-level audit events
• Run-level audit and telemetry measurements
• Node test covering the initial vertical slice

Architecture boundary

The connector exposes source snapshots; the inventory module normalizes them into canonical records. Intelligence modules will consume canonical records rather than Google-specific tables.

Next implementation steps

1. PostgreSQL/Drizzle persistence schema
2. Google OAuth connection and token encryption
3. Concrete Drive API adapter
4. Scan-job orchestration and resumable checkpoints
5. Inventory API and dashboard integration
6. Duplicate, stale, ownership, and permission findings

Validation

The test was added to the repository test command. Local execution could not be completed in the current environment because outbound access to GitHub/package sources is unavailable; CI should run the suite on the branch.