Fpdf2

Added
* [Introducing a rect_clip() function](https://github.com/reingart/pyfpdf/pull/158)
* [Adding support for Contents alt text on Links](https://github.com/reingart/pyfpdf/pull/163)
Modified
* [Making FPDF.output() x100 time faster by using a bytearray buffer](https://github.com/reingart/pyfpdf/pull/164)
* Fix user's font path ([issue](https://github.com/reingart/pyfpdf/issues/166) [PR](https://github.com/py-pdf/fpdf2/pull/14))
Deprecated
* [Deprecating .rotate() and introducing .rotation() context manager](https://github.com/reingart/pyfpdf/pull/161)
Fixed
* [Fixing 159 issue with set_link + adding GitHub Actions pipeline & badges](https://github.com/reingart/pyfpdf/pull/160)
* `User defined path to font is ignored`
Removed
* non-necessary dependency on `numpy`
* support for Python 2

Added
* Python 3.9 is now supported

Added
* new specific exceptions: `FPDFException` & `FPDFPageFormatException`
* tests to increase line coverage in `image_parsing` module
* a test which uses most of the HTML features
Fixed
* handling of fonts by the HTML mixin (weight and style) - thanks `cgfrost`!

Fixed
* images centering - thanks `cgfrost`!
* added missing import statment for `urlopen` in `image_parsing` module
* changed urlopen import from `six` library to maintain python2 compatibility

Added
* Ability to use a `BytesIO` buffer directly. This can simplify loading `matplotlib` plots into the PDF.
Modified
* `load_resource` now return argument if type is `BytesIO`, else load.

Modified
* introduced a dependency to `numpy` to improve performances by replacing pixel regexes in image parsing (s/o pennersr)