We imported some audiobooks from Bibliotheca and discovered that the metadata wrangler incorrectly reports that they are ebooks.

The underlying problem is that metadata_layer.py in core sets Edition.BOOK_MEDIUM as the default medium. Unless we hear something to tell us otherwise, all incoming Editions will be set as ebooks.

In the circ manager, we do hear otherwise, from the license provider. But in the metadata wrangler, we don't hear anything about medium from OCLC or Content Cafe or the other data sources we consult, so the default stands.

The only exception is Overdrive. The metadata wrangler has access to the Overdrive API and correctly records the .medium for the Overdrive titles it's asked about.

There should be no default medium – if we never hear anything about a book's medium (this will only happen on the metadata wrangler) we shouldn't make any assumptions.

Also, we should take what we hear from the metadata wrangler a little less seriously. The cover image is only time we want the metadata wrangler's data to actually replace the data from the license vendor (as opposed to augmenting it).