lanl/SEPIA: v1.1

This is the stable, essentially complete version of SEPIA. We anticipate only minor bugfixes or small feature additions in the future.

Since v1.0.0, we have added:

• To set up a model, use SepiaModel(data) (as opposed to setup_model() function from first release)
• Prediction code finalized, now takes x and t in native (untransformed) space and handles transformations internally
• Added Kronecker-separable design awareness and capability
• Added ability to pass theta constraint functions to SepiaModel
• Added ability to use categorical variables in x and t
• Added more thorough testing
• Fixed some unexpected behavior in model.get_samples()
• Added get_samples() methods for SepiaHierarchicalThetaModels/SepiaSharedThetaModels
• Added to examples (in particular, added shared/hierarchical model example and parallel chains example)
• Simplified install directions and added dependencies in setup.py
• Sensitivity analysis finished and tested against Matlab GPMSA version
• Added to documentation