Imbrium

- added Tensorboard support
- changed show_performance plot to show loss and metric values
- added optional dropout and regularization layers to architectures

- removed batch_size parameter from fit_model method
- hyperparameter optimization added via the Optuna library

- imbrium supports now:
- python 3.7, 3.8, 3.9, 3.10

- added tests for new utils module
- name changes of predictor classes:
- univarstandard => univarpure
- BasicMultStepUniVar => PureUni
- univarhybrid => univarhybrid (unchanged)
- HybridMultStepUniVar => HybridUni
- multivarstandard => multivarpure
- BasicMultSTepMultVar => PureMulti
- multivarhybrid => multivarhybrid (unchanged)
- HybridMultStepMultVar => HybridMulti
- tox added
- outsourced Binder demo notebook to https://github.com/maxmekiska/ImbriumTesting-Demo
- new README.md

- major refactoring of code base
- minor name change from Imbrium to imbrium

- improved source code format
- improved scaling selector mechanism