This PR introduces the plotting of the detector's bounding boxes, if the architecture is Top-Down.

In particular, the bounding boxes are:

• Saved in the *_full.pickle file during video analysis
• Plotted by evaluate_network() when passing argument plotting=True
• Plotted by create_video_with_all_detections(), unless passing argument plot_bboxes=False
• Plotted by create_labeled_video(), unless passing argument plot_bboxes=False
• Plotted by video_inference_superanimal() , unless passing argument plot_bboxes=False
• Plotted by visualize_predictions() , unless passing argument plot_bboxes=False