- (..., 3, 3) full diffusion tensors
- (..., 6) unique components
- tuple of eigenvectors & eigenvalues

There should be full test coverage of these new functions,
but coverage.py is currently not working for me, so not 100% sure.

multiple prefixes

sorted eigenvalues. I do not understand numpy indexing.

dtifit.py. Added new function, decomposeTensorMatrix, which turns a 6
volume tensor matrix image into eigenvalue and eigenvector images.