Merge multispectral shepp-logan phantom and analytical data generation code
The data generation code (createTestStructures and createDefaultInputs) which create the parabolic absorber distributions and corresponding analytical data needs to be refactored to work well with the model refactor (See #30 ) and new testing framework.
Similarly, we already have the code for generating the multispectral Shepp-Logan phantom, which uses ellipses instead of parabolic absorbers. The principle is the same, however.
Combine the two so that a parameter object (describing the distribution) and an image-space object (describing the image coordinates/scale) are input to a process to create an image frame. At the same time, use the same parameter object and a data-space object (describing the data coordinates/scale) to create the corresponding distribution.
This will work for the parabolic absorbers just fine, but I haven't yet run the math on getting the analytical signal that results from a given ellipse with arbitrary orientation and major/minor axis length