A casual glance at the abstract leads me to read the paper's conclusion more as "Teachers who have easy classes and teach to the test provide worse foundations and get better evaluations." This seems like a pretty likely hypothesis that would explain some of the correlation. Some evidence could be gathered for it from ratemyprofs.

I'll read it further when I have time to check for things like linear regression.

ETA: that study looks really good. I am curious with how the data would be effected if students consciously rated easiness separately.