It sounds like you're doing HTML prototyping with Javascript for the functionality iterations (Backbone.js wheee!) so let me approach it from that point of view.
First off, doing things in HTML allows you to focus on the right details of the user experience, whereas with a tool like Axure or paper prototyping you're really zooming out too much to a conceptual level. That's not the right level of detail for iterating on the finer details of how something works. You also can't do things you're trying to do now with those approaches (like data persistency), whereas with HTML and Javascript it's trivial.
I think you're right in the sweet spot in terms of iterating with users and using a quick HTML/JS prototype to change things up based on feedback. That's great. But I'm sure you're bumping into some productivity issues as well, like how to quickly and easily get your updated prototype in front of the test group.
We faced the same problems, so we developed a tool to solve it. Instead of working locally and then committing things to a Git repo or something and having to publish it somewhere, our tool is web-based and contains a HTML/CSS/JS editor in the browser. As a result we now just work from there and everything we save is live instantly.
This is been really significant in terms of how it's changed our approach to prototyping. Now we can be talking to a client on the phone, change something, hit save, and ask them to refresh their browser to see what happened. And we can group usability tests into a few hours spanning several iterations rather than having to invite participants across multiple days. It's a huge efficiency boost.
I realise this sounds very product pitchy but I really believe that when you're doing HTML prototyping our approach is the way to go. It's called Handcraft. Let me know if it works for you.