As we are not discussing DeDRM so much as we are discussing the generics of the mobipocket database, it is fine to discuss it here.
If you like, you can see the gory details in our Wiki:
MOBI
The simple version is that the metadata calibre changes is in the exth headers.
DRM only seals the text blocks (it doesn't even encrypt the images), which is why calibre can extract metadata even from a DRMed book. Once you (try to) convert it, suddenly you realize it is DRMed!
As for Baen, they *might* be using calibre to deliver dual-MOBI (a *.mobi with moth mobi7 and KF8 content, for gracefully falling back on devices that don't support AZW3).
Either way, convert it with calibre, and whatever the problem is will probably be fixed.
Personally, I would use the
EpubSplit plugin to divide those omnibuses into individual books.