Implementation of 2 tools is necessary:
1) Given input volume and output filename (and some magnitude) apply NRR warps to the volume and save it.
2) Calculate trajectories of two warp lists and get mapping of points. Compare this with another trajectory or image and calculate a measure like SSD to get a measure of 'distance' from the correct warp.
Also to experiment with:
1) Get res_info (and later MI) working in 3-D.
2) Try to see how high a tolerance can be used to get equally good
results. This can lead to a speed-up.