FlexTemplate, markdown in multi_cell & bug fixes

Added

• Template() has gained a more flexible cousin FlexTemplate(), thanks to @gmischler
• markdown support in multi_cell(), thanks to Yeshi Namkhai
• base 64 images can now be provided to FPDF.image, thanks to @MWhatsUp
• documentation on how to generate datamatrix barcodes using the pystrich lib: documentation section,
  thanks to @MWhatsUp
• write_html: headings (<h1>, <h2>...) relative sizes can now be configured through an optional heading_sizes parameter
• a subclass of HTML2FPDF can now easily be used by setting FPDF.HTML2FPDF_CLASS,
  cf. documentation

Fixed

• Template: split_multicell() will not write spurious font data to the target document anymore, thanks to @gmischler
• Template: rotation now should work correctly in all situations, thanks to @gmischler
• write_html: headings (<h1>, <h2>...) can now contain non-ASCII characters without triggering a UnicodeEncodeError
• Template: CSV column types are now safely parsed, thanks to @gmischler
• cell(..., markdown=True) "leaked" its final style (bold / italics / underline) onto the following cells

Changed

• write_html: the line height of headings (<h1>, <h2>...) is now properly scaled with its font size
• some FPDF methods should not be used inside a rotation context, or things can get broken.
  This is now forbidden: an exception is now raised in those cases.

Deprecated

• Template: code39 barcode input field names changed from x/y/w/h to x1/y1/y2/size