Add ability to fit with beam + noise components

[e-koch]

Large spatial frequencies will be dominated by the beam and noise. Add the choice to add these components for the fits of the 2D power-spectrum methods.

Since the beam would have to be specified, the fitted data would be transformed before fitting the model. The noise model, as an additive term before taking the log, cannot be written as a linear model anymore. That's an issue for the segmented linear model.