Retry downloading images

Currently, only the images that have not been downloaded are downloaded when the command is restarted. But when we have a lot of data, the whole part about getting the transcriptions has to be done again, but it can be quite long.

Downloading Hugin-Munin v3 takes about 1 hour (not including downloading the images), but with over 12k paragraphs to download, many images are in error and it's very time-consuming to have to restart the command each time.

I attach the log of last time I re-launched it: log.txt