Investigate the edge pixel in the direct method

[DanHickstein]

Following the discussion is #214 (comment), we may want to think about how to deal with the last element of the output. Currently, this is strictly zero, which makes some degree of sense, because a basic assumption of the Abel transform is that the intensity goes to zero at the last pixel. However, other transform methods (at least the hansenlaw) allow this pixel to have non-zero values. Potentially we could pad the array by an additional value (either zero, or the same as the last pixel) and then trim the final array.