[v3gas]

This currently returns an error if it doesn't find exactly one table in the image, so it might be able to work with larger images, but probably not if there are multiple distinct blocks of text.