Aside from pissing off Twitter, and possibly getting a cease and desist, what's preventing developers from building a translator that sits between Twitter's web layer and their native application client? Couldn't something be developed that loads Twitter into a hidden webview that's locally scraped for the purpose of re-display however the developer pleases on their client? This implementation wouldn't require the tokens, and wouldn't be constrained by their arbitrary limits.
You answered your own question. People seem to be under the impression that just because something is accessible with a web browser it no longer falls under copyright and you can use it how you want. This has never been true. Any web site can make a terms of use (and most do) that prohibit you scraping their data.
So what about user style-sheets? They modify the appearance of the data, but I've never seen anyone sued over it. Nobody sues IE for rendering HTML differently. In this case it'd just be another 'browser'.
But I suppose they'd just wind up getting sued.