Add an option `--recognise` to start with manuscript images ? To do, - recognition ; - segmentation in words ; - abreviation expansion. To be able to go from manuscript image to collated text.