There's no point in having clean_canonical_model, the canonical names will always be clean. It's almost surely not the case this doesn't do anything as most cameras don't have the issue in #19. But feel free to remove the camera you personally have as you can test it. It's to do the mass delete that I want to have the regression testing in place.

Re: clean_canonical_model:

In the case of a Canon EOS DIGITAL REBEL T3i, if there isn't a clean canonical model, should the canonical_model be Canon EOS 600D or EOS 600D?

Currently in this PR:
canonical_model = Canon EOS 600D
clean_canonical_model = EOS 600D

Do you need both for some reason? I'd use clean_model of the parent camera as canonical_model of itself and all aliases. If the point is to have one way to index into a specific camera having two versions of a "canonical" name sounds very strange. And using the clean one is best as cameras often have wonky names and the clean ones unify a lot of stuff so they're nicer if you want to automatically match something.