Stochastic spectrogram object

I've added two new objects, omega_spectrogram (subclass of gwpy.spectrogram) and omega_spectrum (subclass of gwpy.FrequencySeries), to better handle the setting of alpha, fref, and h0 in stochastic run results. these have a built-in call to the spectral.reweight function which comes in handy.

In the process we've also moved around some functions, and fixed bugs here and there to do with the spectral re-weighting functions.

I also made point estimate and sigma attributes into settable attributes so that they are easier to track. We may want to put in more errors/warnings into the setters.