Draft: We need to store reference images of all source pdf pages so that we can do rectangle extraction

• Build model to hold image info
• Tweak mocker for use for building PDFs with reference QR codes
• extract images from the mocked pdfs on source upload
• extract locations of qr-codes from those mocked images
• add JS to select a rectangle from an image.
• add new django-app thingy for rectangle selection
• manager command for doing this.
• clean up, add seatbelts and error handling
• add zipfile download for extracted rectangles.
• clean up / remove the old scan/image_proc code... also give appropriate credit in the new code.
• profit.