Text this: Compressing pathology whole-slide images using a human and model observer evaluation