Experiment with alternate alpha masking functions

[perrygeo]

Others to consider:

• simple masking but with a threshold
• random walker (start from nodata and move outward)
• sieve
• thresold masking > shapes then select shapes on edge
• using majority filter/moving window

[perrygeo]

It would be nice to have a way to run test datasets on all available mask funcs in order to visually inspect.