Pastastore

In this release:
- Improved example notebooks (specially for users coming from pastas.Projects
- Improved naming timeseries logic (removed support for multi-column DataFrames)
- Improved consistency between Connector objects.
- Add `get_parameters()` and `get_statistics()` methods to PastaStore.
- Bug fix for storing and loading from zipfiles.
- Add checks when adding models and improve error messages if stuff goes wrong.
- Support for pastas 0.15.0 (refactor of internal methods)

New functionality:
- save store content to zip-file (a simple method to share data) and load data into PastaStore from zipfile (3 and 9, 10): `Pastastore.to_zip()` and `PastaStore.from_zip()`
- added `PastaStore.get_parameters()` and `PastaStore.get_statistics()` methods (Fixes 3)
- Allow `pastas.Models` to be read and added to databases in dictionary/json form. When the functionality of the pastas Model is not required, it is faster to access the models as dictionaries. This update allows the user to handle models in the PastaStore this way.
- When adding a model based on timeseries that are not already in the PastaStore, the oseries and stresses are added to the PastaStore automatically and a warning is shown.

Improvements:
- `DictConnector` now explained in readme (8)
- Connector methods (e.g. `store.conn.get_oseries()` are now available directly from the PastaStore object, i.e. `store.get_oseries()`.
- formatted docstrings with docformatter
- improve docker-compose file for using MongoDB
- and quite a few minor improvements...

Fix to `pastastore.get_tmin_tmax()`

First release of the pastastore module.