A deep attention-assisted and memory-augmented temporal convolutional network based model for rapid lithium-ion battery remaining useful life predictions with limited data