Compressive Differentially-Private Federated Learning Through Universal Vector Quantization