The rich text is not stored for each card, so it is not possible to implement this request reasonably. Storing the rich text with every card would be a disaster. Consider for example a 100k seal image stored with every card in King County. That would be 4000 * 100k or 400MB of data, vs 100k of data currently.
Storing the text with each card would also eliminate (or at least complicate) the very usefull feature of being able to change the text on one card and have all other cards reflect the change.