support for frequency-dependent calibration files

[cparcerisas]

The idea right now for the packages is that a flat value calibration is used during processing. This is done because the conversion to uPa is done in the time domain (directly on the raw signal) and to apply a frequency-dependent calibration one would need to compute the spectrogram, calibrate, and then re-compute the signal.
So far, how we're dealing with it is using the flat calibration value for all the operations and then apply the frequency-dependent calibration to the frequency-dependent results. This can be done as shown in this example (pyhydrophone computes for you the difference between flat calibration and freq-dependent calibration) :
https://github.com/lifewatch/pyhydrophone/blob/main/examples/plot_freq_calibration.py

Implementing it in pypocc would then be applied during the classification of the clicks, where a spectrum is computed (not during the detection of all the possible clicks).