We should be able to get all code blocks within markdown, ideally grouped by language and in the order found