HTMLMixin not needed anymore - more HTML tags supported - demonstration Jupyter notebook

Added

• demonstration Jupyter notebook: tutorial/notebook.ipynb
• new .default_page_dimensions property on FPDF instances
• support for description list (<dl>), description titles (<dt>), description details (<dd>) and code blocks (<code></code>) in write_html() - thanks to @yk-jp & @seanpmulholland
• support for monochromatic images (PIL image.mode == '1') thanks to @GerardoAllende
• the 1000+ unit tests suite is now executed under Linux and Windows, with extra timing & memory usage checks ensuring we control fpdf2 resource usage
• new translation of the tutorial in עברית, thanks to @TzviGreenfeld
• new documentation for using PyPDF2 with fpdf2, added by @devdev29: https://pyfpdf.github.io/fpdf2/CombineWithPyPDF2.html
• new documentation for using Jinja with fpdf2: https://pyfpdf.github.io/fpdf2/TemplatingWithJinja.html

Deprecated

• HTMLMixin is deprecated, and not needed anymore: the write_html() method is now natively available in the FPDF class - thanks to @yk-jp

Removed

• open() & close() methods, that were only used internally and should never have been called by end-user code
• FPDF.state, which was an instance of the DocumentState enum, and has been replaced by moving the final rendering logic into a new fpdf.output module

Fixed

• after an "empty" cell(), ln() applied a line height of zero #601
• when using multi_cell() with max_line_height to render multiline text, the last line is now rendered like all the others
• templates don't leak graphics state changes to their surroundings anymore; #570
• automatic page break is never performed on an empty page (when the Y position is at the top margin)
• fixed insert_toc_placeholder() usage with footer() and {{nb}}; #548
• the SVG parser now accepts stroke-width attribute values with an explicit unit, thanks to @gmischler; #526
• the SVG parser now accepts absolute units for width and height attributes, thanks to @darioackermann; #555
• write_html() method now correctly handles whitespace when parsing HTML. <pre></pre> blocks still maintain spaces, tabs and line breaks.

Changed

• the first parameter of FPDF.add_font() is now optional: if it is not provided, the base name of the fname font path is used to define the font family. Hence pdf.add_font(fname="fonts/NotoSansArabic.ttf") will define a font named NotoSansArabic.
• the output of embed_file() is now a PDFEmbeddedFile, not a string, but the internal file name can be retrieved through its .basename property
• forbid use of get_y() & local_context() inside unbreakable() as it is currently not supported; #557
• fontTools minimal version requirement set to 4.34.0; #524