single threshold vs adaptive threshold

[MatthewWhiting]

This is related to #59.

It would be good to have (IWBGTH) the ability to select either a single threshold (this is the current functionality) or an adaptive threshold.

The latter case could either be:

• a S/N cut depending on the noise in a given 2D image
• a S/N cut depending on the noise in a given 1D spectrum
• a spatial filter (in the style of the ASKAPsoft continuum case) to determine the local noise
• a spectral equivalent
• a "3D" equivalent, looking at a bunch of channels over a small range of frequencies.

Where the stats are calculated would depend on the single/adaptive nature of the threshold, and on what the adaptive case was.