Release 2.4.0

Bump to version 2.4.0 due to incompatible API changes in mixed_inv_fft, mixed_fwd_fft, and mixed_make_segment_convolution to use uint32_t for the framesize instead of uint16_t, as now larger framesizes for FFT are supported.

spiralfft has been removed in favour of pffft, which still has acceptable performance, supports larger FFT windows, and is much smaller in binary size. Ridiculously so, in fact.

Release 2.3.1

Version bump to 2.3.1

Release 2.3

Forgot to bump version file... sigh.

Release 2.2

Document the convolution segment

Release 2.1

Try a build action