Remove ML model artifacts from git and add FCM push notifications

[Copilot]

The repository was bloated with 100MB+ of ML model files, and the system lacked real-time alerting for malicious requests.

Changes

Artifact Management

• Removed ML models (.h5, .tflite) and tokenizers (.tokens) from git tracking
• Added GitHub Actions workflows:
  • build-models.yml: Builds models on dataset/script changes, stores as artifacts (30-day retention)
  • clean-artifacts.yml: Daily cleanup keeping only latest artifacts
• Updated .gitignore to prevent future commits of build artifacts

Push Notifications

• Implemented FCM integration via notification_service.py (async, aiohttp)
• Integrated into both Dev and Prod server_socket_main.py:

  if final_result == "MALICIOUS":
      asyncio.create_task(
          notification_service.notify_malicious_request(
              str(addr), uri, uri_final_result, body_final_result
          )
      )

• Configuration via FCM_SERVER_KEY env var and device_tokens.json
• Graceful degradation when disabled (no FCM key or device tokens)

Documentation

• SETUP_GUIDE.md: Firebase setup, device token configuration
• QUICK_REFERENCE.md: Common operations
• .env.example and device_tokens.json.example: Configuration templates

Testing

Unit tests in test_notification_service.py validate service initialization, notification sending, and malicious request filtering.

Original prompt

I want you to clean the actions artefacts once the file is added to the repository.

Also I didn't received a push notification, I want to receive a push notification on the app

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.