yep, that's good
I need more time :) I will try to do some after I have the current reports out of the way.
@jerome The tests are for comparing HTML output and screenshots of PDF output. And yes, it should be exact.
Jean-Paul is very sensitive to changes in the layout of our PDF docs, so I prefer to leave everything as it is and let the tests pass.
The html neither parsed nor reassembled, this is a pandora's box = there is a million things being done with the HTML to get it into wkhtmltopdf and to not get unknown error but a clean PDF. If you want to change things including using a parser, feel free, but I surely don't have to time to rewrite all the exports (book, slideshow, press release, letters, etc + 10 reports) and then test equally long to get back to the current state :)