There is currently no automated testing, not even an attempt at checking we get a PDF generated from some example html. This needs to change.