[Chore] Migrate shared/database/ to penguin-dal

[PenguinzTech]

User Story

As a maintainer, I want to replace the dual PyDAL+SQLAlchemy database pattern with penguin-dal so that schema is defined once and the codebase follows penguin-libs standards.

Background

Elder currently uses a custom shared/database/ module (~4 files) implementing:

• init_db(app) — PyDAL initialization with migrate=False
• init_sqlalchemy_tables(app) — SQLAlchemy create_all()
• get_database_url(app) — URL normalization (postgresql:// vs postgres://)
• DatabaseManager — Read/write splitting with replica support
• create_db_connection() — Factory with retry logic
• ensure_database_ready() — Startup connectivity check

penguin-dal (pip install penguin-dal) replaces the entire dual-library pattern:

• Single table definition (no separate SQLAlchemy models + PyDAL define_table calls)
• Built-in connection pooling, multi-DB, async support
• Flask integration via init_dal() / get_db()
• PyDAL-identical query syntax

Per database.md: "Migrate at the next major version bump — not mid-release."

Acceptance Criteria

• penguin-dal replaces PyDAL + SQLAlchemy dual pattern
• All table definitions consolidated (single source of truth)
• shared/database/ module removed or reduced to Elder-specific wrappers
• Read/write splitting preserved (DatabaseManager equivalent)
• All existing API endpoints work without query changes
• migrate=False equivalent enforced (Alembic still manages migrations)
• Tests pass (unit + integration)
• Linting passes
• Security scan passes

Notes

• This is the highest-effort migration item — database layer changes carry significant risk
• Requires penguin-dal to be stable (currently v0.1.0)
• Must test all 4 supported databases: PostgreSQL, MySQL, MariaDB Galera, SQLite
• shared/models/pydal_models.py table definitions will need rewriting