py-pdf/fpdf2 2.8.4 on GitHub

documentation on internal linking with variable page numbers
documentation on using the Ibis library
clarified docstring for arc() method to document x and y arguments (#1473)
made numpy an optional dependency to improve the performance of pack_codes_into_bytes() - cf. issue #1380
macOS is now included in the GitHub Actions CI pipeline to ensure cross-platform test coverage

FPDF.write_html(): Fixed custom styling of <p> & tables - issue #1453
an IndexError was raised in some cases when rendering HTML with nested lists - cf. issue #1358
an IndexError was raised in some cases when using text_columns() with text shaping enabled - cf. issue #1439
handled case of a smaller last column when using text_columns() - cf. issue #1442
minor regular-expression optimization in the SVG parsing module, nullifying any risk of ReDOS attack
a performance issue when using FPDF.offset_rendering() or FPDF.unbreakable() with large fonts - cf. issue #1444
slightly invalid PDF files were produced since v2.8.3, due to path construction commands being inserted inside BT/ET contexts when using underline or strikethrough text - cf. issue #1456
multi_cell() text clipping bug - issue #1471
HTML lists bullets are now properly displayed when using unicode fonts - cf. issue #1496
clarified documentation in Maths.md regarding DataFrame string conversion for PDF rendering
children of <a> tags in SVGs are now correctly rendered - cf. PR #1522

py-pdf/fpdf2 2.8.4 Dropping support for Python 3.8 - automated tests run on MacOS - several bugfixes on GitHub