One thing I noticed after resurrecting the test suite: text in double tics is now rendered as "<i>" instead of "<em>". Was this a consciously-made change, or just an error introduced into the parser at some point?