The 'labelled image' in the labelling_tool were keeping labels and lists of completed tasks alongside any
other metadata and attemting to keep it all sane. The code to do this was getting rather out of hand,
so WrappedImageLabels keeps the labels and associated metadata together and switches between JSON and
Python object based representations of the labels as needed.
Refactor: replaced labelled image classes in labelling_tool module with new API in labelled_image module (breaks compatibility)
The 'labelled image' classes were used by the Flask and Qt based labellers for managing images and their associated labels. File and memory based images and labels were supported. The API was a mess hence the re-design. This should make it easier for others to extend this class hierarchy and make use of it.
The functionality of 'labelled images' has now been separated into the ImageSource class hierarchy that
support image access and the LabelsStore class hierarchy that supports retrieving and updating labels.
The AbstractLabelledImage, InMemoryLabelledImage PersistentLabelledImage and LabelledImageFile
now have deprecated wrapper functions in image_labelling_tool.labelling_tool to allow old code to hopefully work.