Hi,
I decided to test some of my old documents whose content and layout caused issues with older versions my PyMuPDF - to good results at first glance. Thanks for keeping up the great work on this package 
One thing which tripped my pipeline up a smidge, was a possible change in the flags used to mark text extracted from pictures.
So, I wondered if there were any other such tags? I can’t find reference to <!-- Start of picture text --> ... <!-- End of picture text --> markers anywhere in the documentation - perhaps there are other content types marked in a similar manner?
One final observation is that ignore_images (bool) is set to true, so I’m a little surprised any were present?!