What are padding and stride in convolutional neural networks (CNNs) and why are they important? How does padding affect the spatial dimensions of feature maps? How does stride control the movement of the convolutional filter across input data? What impact do different padding and stride values have on model performance and computational efficiency? What are the advantages and limitations of adjusting padding and stride in CNN architectures?