📝 **Task:** Implement two functions: 1. `register_model(model_id, metric)` → returns the new version number (1, 2, 3, ...). Stores each registration in a module-level `REGISTRY` dict keyed by `model_id`. 2. `detect_drift(train_mean, train_std, live_values)` → returns `True` if the mean of `live_values` is more than 2σ away from `train_mean`. Then run the checks shown below. Expected output (5 lines): ``` 1 2 {'version': 2, 'metric': 0.87} False True ``` 📋 Implement the function above. Tests run automatically. 💡 **Hint:** Re-read the theory if you get stuck.