Mdt

====================
This version is significantly faster than previous versions when run using a GPU. All users are recommended to update
to this version.

Other
-----
- Following changes in MOT.
- Small cosmetic improvement in the C code.

====================

Added
-----
- Adds NODDIDA.
- Adds method argument to the mdt sample function.

Other
-----
- Removed redundant super arguments.
- Refactorings following changes in MOT.

====================
- Removed some non-ascii characters for compatibility.

====================
- Following changes in MOT, in particular how the optimization routines are called.

====================

Changed
-------
- Updated makefile to use twine for uploading to PyPi.
- Replaced Grako for Tatsu, as Grako was no longer supported.
- Removed the Tatsu debian package and added it as a Pip requirement.
- Removed six as compatibility layer.

====================

Added
-----
- Adds documentation on debugging OpenCL elements.
- Adds a button to the maps visualizer to only show the set options in the textual frame.
- Adds simple data compression to the gradient deviation computations in the case of zeros off the diagonal.
- Added the covariance terms to the error propagation of Tensor FA.

Changed
-------
- Changed method signature of saving view map plots.
- Small update to the unweighted volume computation in the Protocol, it now multiplies the gradient vector with the diffusivities to account for non-normalized gradients.