Audiotoolbox

- Made filterbanks indexable to gain excess to individual filters
- fixed a small bug in the erb to 3db calculation function of the gammatone
filter which could result in a type error

- Added a default auditory gammatone filterbank
- Added a default fractional octave filterbank
- Added function to convert frequency into band number following ANSI norm
- Fixed a small bug in _copy_to_dim which would remove the last dimension of an error if it equaled 1
- Deleted broken audio playback functionality
- Moved filterbanks from audiotools.filter to audiotools.filter.bank submodule
- audiotools.freqarange now supports octave spacing

- Fixed a bug in the FrequencyDomainSignal.to_timedomain() method

- Added parameter to auditools.dbfs to specify if dB fullscale is relative to peak or rms level
- Deleted the long depricated Signal.add_noise_noise method
- Implemented Signal.stats submodule
- Added deprecationwarning to Signal.calc_dbfs and Signal.calc_dbspl which
moved to the Signal.stats submodule
- Added the option to directly apply filters when generating partly correlated noise
- Added the audiotools.cmplx_corr function which calculates the
complex-valued correlation coefficent
- Renamed calc_coherence to cmplx_crosscorr
- Fixed a bug in signal.phase_shift that sometimes resulted in a complex valued signal

- Fixed the shape of the channels after appling a filterbank
- Added a summary method that prints information about size and shape of the object