Enviar Registro por Correo electrónico: Learning Low-Precision Structured Subnetworks Using Joint Layerwise Channel Pruning and Uniform Quantization