This is fallout from https://jira.nypl.org/browse/SIMPLY-1306. For a long time, the metadata wrangler has assumed that every non-Overdrive title it's ever heard of is an ebook. Most of the time this assumption is right, but when it was wrong it caused big problems.

The big problems have been fixed but the bad data remains on the metadata wrangler and it would be better if the metadata wrangler didn't provide bad data, even if that data gets ignored by the client.

The simplest way to do this would be to remove all media type information from every non-Overdrive Edition and recalculate all the Work presentations.

However, 'recalculate all the Work presentations' on the metadata wrangler is a really big task, even if we ignore the Overdrive and Gutenberg books.

The number of books affected is relatively small. Unfortunately I don't have an easy way of determining exactly which books are affected.