feat: Implement multi-database architecture

This commit introduces a three-database architecture to the application,
as specified in the `db_logic.md` file.

The changes include:
- A SQL initialization script (`db/init/init.sql`) to create the
  `USERS_ALMA`, `vanity_hr`, and `vanity_attendance` databases and their
  respective tables.
- SQLAlchemy models for all tables, organized into separate files
  within the `models` directory.
- Refactoring of the database connection logic in `modules/database.py`
  to support connections to all three databases.
- Creation of a `modules/logger.py` to handle request logging to the
  `USERS_ALMA` database.
- Updates to `docker-compose.yml` to mount the initialization script and
  build the bot image locally.
- Updates to `.env.example` to include the new database environment
  variables.
- Restoration of the data dictionary to `db_tasks.md`.

modules/rh_requests.py

@@ -5,7 +5,7 @@ import string
 from datetime import datetime, date
 from telegram import ReplyKeyboardMarkup, ReplyKeyboardRemove, Update
 from telegram.ext import CommandHandler, ContextTypes, ConversationHandler, MessageHandler, filters
-from modules.database import log_request
+from modules.logger import log_request
 from modules.ui import main_actions_keyboard
 from modules.ai import classify_reason