Gallica import takes wrong images when more than 10

I noticed when I tried to import a newspaper with 36 pages in it. This line makes it take only the last digit as image_id.

We need more tests on this part and a cleaner way to retrieve the ID (a regex?).