Wsknn

- (bug) `random.sample(sessions)` bug: https://github.com/nokaut/wsknn/issues/64
- (enhancement) check if sessions and items have the same datatypes in both mappings (session-items map and item-sessions map) (https://github.com/nokaut/wsknn/issues/65)
- (enhancement) batch prediction with `batch_predict()` function

- (docs) updated JOSS badge,
- (docs) updated docs,
- (enhancement) updated import paths,
- (feature) save and load model with `joblib`

v1.2.0.joss-publication
This release is linked to the JOSS publication.

* (docs) changed README example,
* (docs) added demo-readme example to demo-notebooks section,
* (docs) demo example in documentation updated,
* (docs) link to the documentation page added to README,
* (enhancement) item-sessions map may be derived from session-items map,
* (feature) data can be read from the flat record structure,
* (feature) data can be parsed from dataframes

- reads and parses CSV files
- refactored parsing functions

The release sets minimal version of Python to 3.8, has updated docs, and CI tests. It is mostly a dirty job to make life of developers and maintainers easier.

- updated session selection algorithm, added `weighted_events` parameter that allows weighting the closest neighbors by the vector of weights,
- new tutorial with a comparison of different recommendation settings,
- updated documentation.