Refactor: DB Layer Standardization & Migration

Issues with current implementation:

1. Schema drift: schema.sql doesn’t reflect what the code actually uses → fragile deploys.
2. Raw SQL everywhere: repeated query/scan logic, inconsistent transactions, harder to test.
3. Unsafe patterns: dynamic SQL (fmt.Sprintf) and ad-hoc casting increase risk.
4. No migration workflow: schema changes aren’t tracked or reproducible.
5. Mixed structure: models, queries, and DB concerns are spread across packages/files.