I can't address the main question, but this part is something I can comment on:
> Surely if people are downloading accross http they can't resume anyway?
Yes, they can, if the download manager keeps track of the last 'piece' of the file successfully downloaded, and saves this information on the client machine. HTTP/1.1 supports 'partial content' requests, so that a client can request, for example, byte 1,032,468 through byte 1,094,762 of a file from the server. If this request is sucessful, you'll see a 206-Partial Content response in your server log file.
See RFC2616 "Range Units [w3.org]" and other sections linked from there.