Prepare next PyAF 4.0 release

[antoinecarme]

This is a placeholder to collect ideas/tasks for making the next 4.0 release available.

1. Python 3.10 support Python 3.10 support #186
2. Add Multiplicative Models/Seasonals Add Multiplicative Models/Seasonals #178
3. Speed Performance Improvements : Internal Dataframes index refactoring #190 , Lag generation process optimization #191
4. Exogenous data support improvements : Model Building Fails when Exogenous Variables are configured but not used (ARX disabled by user) #193, Exogenous data are not used in MLP and LSTM-based models #197 , Make the Modeling Scale when the Number of Exogenous Variables Increases  #198
5. PyAF support for ARM64 Architecture PyAF support for ARM64 Architecture #187
6. PyTorch support : Investigate PyTorch-based LSTM and MLP models #199
7. Improved Logging : Add some messages to reflect the progress of the trainining process #185
8. Bug Fixes : Crash when training using example code - An attempt has been made to start a new process before the current process has finished its bootstrapping phase #156, Overflow in cumprod (relative difference transformation) #179, Croston implementation is slow #182, The signal is not rescaled before applying some transformations  #184
9. Release Process : Pre-release Benchmarks Release Process : Pre-release Benchmarks #194
10. Release Process : Profiling and Warning Hunts Release Process : Profiling and Warning Hunts #195
11. Release Process : Review Existing Docs Release Process : Review Existing Docs #196, Document the Algorithmic Details of PyAF #35

Any short task that can be implemented and tested before 2022-07-14

Release date : 2022-07-14

[antoinecarme]

Done.