Makes sense. Obviously this is better for something being released to the public but during testing I didn't know if (even inefficiently) asv could be used as a one stop shop (a la webrick).
That shouldn't be too hard to hack in, I think... just find some way to capture all /images/* URLs or something and open up a file and send the contents. You might have to be a little careful with MIME types and so forth, but it seems pretty straightforward.
Just take any URL that is not defop'ed to mean a file name.
Currently, if no op is defined, the server responds with "Unknown operator.". Replace that with the code for opening a file, and if this fails, respond with a proper 404 Not Found message.
