Dilatometer algorithm needs refinement

[szmoore]

As noted in the report.

Suggest going back to a seperate test program (but use real images). Perhaps use a real test image but manually add an offset to it to simulate expansion of the can. Then actually plot the recorded pixel position of the edge vs the simulated position. Then refine the algorithm until they match well enough.