fugw 0.1.1

This release includes:

• dense and sparse solvers for the FUGW optimal transport problem in which the marginal constraints are formulated with an L2 norm instead of a KL (set divergence='kl' or divergence='l2' when defining a mapping) #14 #29
• a series of features and examples to work with volumetric MRI images (as opposed to surface MRI images) #19
• enabling users to set the initial transport plan before fitting a mapping #27
• util functions to save mappings and load them with / without coefficients (for faster loading) #18
• tracking the contribution of each component of the loss throughout training #16
• tracking the evolution of the fugw loss on validation features throughout training #20
• using training callbacks, useful to track custom metrics during training for instance #21 #23
• many bugfixes

Thank you @6Ulm @pbarbarant and @bthirion !