Explore predicting/propagating bounding boxing from one image to the next in the selected span.

[LeeWannacott]

Instead of having the user put bounding boxes onto each of the images of the span selected manually. Explore using a tracker to predict where the bounding box on the next images in the selected span should be. Look into the trackers in OpenCV contrib: https://pypi.org/project/opencv-contrib-python/ such as median flow tracker. Relevant file: image_selector_from_video.py