Stftpitchshift

Enhancements:

* Reset vocoder phase buffer on parameter change 45
* Add CMake options to enable optimization flags 41
* Approximate the arctangent function 40
* Use std::span instead of std::vector 39
* Reduce latency with asymmetric windows 38
* Support integer data types in Python implementation 36
* Exception about resizing numpy arrays when debugging (and only when debugging) 31
* Preemptively clear output memory 30
* Add WebAssembly library enhancement 28

Major API changes:

* C++ 20 instead of 11
* `std::span` instead of `std::vector`

Follow-up project:

* [stftPitchShiftPlugin](https://github.com/jurihock/stftPitchShiftPlugin)

* Create an [audio demo website](http://jurihock.de/stftPitchShiftDemo) #26
* Add timbre shifting feature in addition to the pitch shifting 25
* Refactor linear resampling routine 21
* Refactor cepstral liftering 11

* Allow semitone and cent input alternative to pitch shifting factors 20
* Add optional spectral RMS normalization to avoid output amplitude clipping 10
* Add homebrew formulae `stftpitchshift` and corresponding tap [jurihock/stftpitchshift](https://github.com/jurihock/homebrew-stftpitchshift) #19

* Add support for custom FFT backends 6
* Add new default FFT backend 7
* Add real-time processing interface 18
* Add double precision support 8
* Add CMake `find_package` support in the VCPKG build 13
* Fix negative spike in output file on Windows 4