Probeye

Added
- The Scipy-solver was equipped with a maximum a-posteriori estimation method.

Removed
- The 'knowledge_graph_import.py' script was removed. The functionality was moved to the integration test 'test_query_results.py'.

Added
- Added new submodule 'ontology' for an import/export of the problem's knowledge graph
- Added an ontology file 'parameter_estimation_ontology.owl'

Changed
- The forward model's 'definition'-method was renamed to 'interface'.
- The user-definition of the forward model's 'interface'-method is now mandatory.
- The prior-parameters of the normal and lognormal distribution have been renamed from 'loc', 'scale' to 'mean', 'std'.
- The pyro-solver has (and its dependencies have) been removed from the package. All solvers are now based on numpy-arrays.
- Fixed a bug resulting from an update of numpy causing an error when using dynesty with a given seed.
- Removed method InverseProblem.transform_experimental_data since it is no longer necessary without the pyro/torch-solver.
- The specification of a parameter's type (model, prior or likelihood) is no longer necessary (it can be detected automatically).
- The multivariate normal prior is now an own prior type and not anymore included in the normal prior class.
- Updated the docs (however, still work to be done).

Changed
- The main class InferenceProblem was renamed to InverseProblem.

Changed
- Before the evaluation of a likelihood function, the parameter domains are checked.
Added
- Latent parameters have domains now. A respective class has been added to the parameters.py-file.
- More tests have been added.