[ENH] Alternative spectral model

[ryanhammonds]

The PR here describes a more accurate timescale PSD model relative to the Lorentzian function. This function was derived out from the fft of an exponentially decaying ACF. In practice, this alt model is going to be equivalent to a Lorentzian if the upper freq bound << nyquist. This should be added as an option in this package at some point.