Implementation and derivation of "Variational Bayesian inference for a nonlinear forward model." [Chappell et al. 2008] for arbitrary, user-defined model errors.
Thank you for implementing that!
A simple test case -- e.g. comparing a concatenated multi_me.evaluate with a multi_me.__call__ -- would be nice though.
Add an evaluate routine to the multi model error that stores a dict of single model errors.