Eds-scikit

Changed
- Support for pyarrow > 0.17.0

Added
- biology module refacto
- load_koalas() not by default in __init__.py but called in the improve_performance function
- adding app_name in improve_performances to facilitate app monitoring

Fixed
- Generation of an inclusion/exclusion flowchart in plotting
- improve_performance moved from __init__.py to io/improve_performance.py file
- Caching in spark instead of koalas to improve speed

Added
- Module ``event_sequences`` to visualize individual sequences of events.
- Module ``age_pyramid`` to quickly visualize the age and gender distributions in a cohort.
Fixed
- Compatibility with [EDS-TeVa](https://github.com/aphp/edsteva) and [EDSNLP](https://github.com/aphp/edsnlp).

Added

- BaseData class as a parent class for HiveData, PandasData and PostgresData.
- Phentyping class with 4 implemented phenotyes.
- Custom logger to display useful information during computation.

Fixed

- Add caching to speedup computations.
- Updated method to persist tables as parquet locally, with a support for ORC-stored I2B2 database.

Added

- Allow saving DB locally in client or cluster mode.
- Add data cleaning function to handle incorrect datetime in spark.
- Filter biology config on care site.
- Adding person-dependent `datetime_ref` to `plot_age_pyramid`.

Fixed

- Consultations date for OMOP & I2B2

Added

- New BackendDispatcher to handle framework-specific functions
- I2B2 to OMOP connector

Added

- Adding CITATION.cff
- Using `mike` as a documentation provider

Fixed

- Correct build to PyPI
- Renaming from `EDS-Scikit` to `eds-scikit`