[ICCV'23] VQD-SR: Learning Data-Driven Vector-Quantized Degradation Model for Animation Video Super-Resolution