Get loads of those, all from MS IPs associated with msnbot and all using more or less standard browser UAs. Been going on for months.
They're a pain but possibly serve a purpose. Google and Yahoo do something similar. I assume something has to provide genuine cache images of a site and need to use a proper browser to grab css etc in context.
Alternatively, since they've refined the refer from the original totally irrelevant garbage to something approximating the site's content perhaps there is some index refining going on. Or, being MS, perhaps they've just let the apprentices loose in the playground. :)