Isn't the element "id" supposed to be unique within an xml document? Having two items with id="cover" is probably technically invalid. (Not that any two xml parsers will ever agree on anything.) I realised that when I ran across this in one of my test books...
Code:
<item id="cover" href="cover.xhtml" media-type="application/xhtml+xml"/>
I tried to tweak that to use a different id (and add the id="cover" element pointing to the jpeg) and only succeeded in creating something that the nook would recognise the cover of, but couldn't open the book itself.
Could this all be as simple as perhaps the nook doesn't realize that 'jpeg' has in 'e' in it?