Stretch: L2 Regularization (Ridge)

Implement:

def ridge_grad(
    input: np.ndarray,
    weight: np.ndarray,
    bias: np.ndarray | None,
    target: np.ndarray,
    lam: float,
) -> tuple[np.ndarray, np.ndarray | None]:
    """
    Compute the gradient of the MSE loss with L2 regularization.
    """

Your function should:

1. Compute the base MSE gradient using 03-mse-gradient's mse_gradient.
2. Extend just the weight gradient with the additional penalty term derived in Theory, one that grows with both the weight's own magnitude and lam.
3. Leave the bias gradient untouched, bias is never regularized, including the bias is None case.