Replying to rmccue:
[...] As far as I can see, there's no point in adding an upstream patch to SP. I'm fairly sure that the RSS specification, somewhere, mentions that a correct MIME type must be used, though I can't find this.
Well SP actually handles the Feed. What happens here is that when the URL is requested for the first time and (only) some mimetypes are not found, SP assumes that it must be a website and tries to autodiscover. In this case it fails because it is actually a feed.
for a real improvement SP should make the autodiscover that clever if checking a document for further links to feeds to actually think about to check if the whole document already is a feed.
a workaround for a user might be to not pass the feed url but the website that links the feed. unfourtionatly with the feed URL provided i cannot test.