If an image URL contains a non-ASCII character (e.g. `kartöffelchen.jpg`), the html returned by the markdown parser contains the encoded path `kart%C3%B6ffelchen.jpg`. We must decode the path before checking if the image file exists.
