Scalecast

Added
- Added `Forecaster.synthesize_models()`
Changed
Fixed
- Made compatible with more recent versions of the shap and xgboost packages

Added
Changed
Fixed
- Fixed `notebook.results_vis()` function when the dictionary keys are not strings.
- Transfer learning no longer fails when fitted values cannot be generated -- still investigating the root cause of this issue.

Added
- Added multivariate transfer learning
Changed
- `must_keep` in `Forecaster.auto_Xvar_select` can now be str type and will maintain the expected behavior.
Fixed

Added
Changed
Fixed
- Fixed how ARIMA model was reading future xreg values.

Added
- Added `fi_try_order` as an argument in `tune_test_forecast()`.
Changed
Fixed
- Got rid of the `AdditiveExplainer` and `GPUTreeExplainer` in `Forecaster.save_feature_importance()` which do not work with feature importance as it is currently set up.

Added
- Added more feature importance options, all sourced through the shap library.
Changed
- shap is now a requirement and eli5 is not.
- Changed `Forecaster.reduce_Xvars()` to use only shap feature importance to rank features.
- Removed `fi_method` argument from `tune_test_forecast()`.
Fixed
- Fixed how a pandas function was called that was raising a warning.
- Fixed feature importance to use shap only with TreeExplainer, PermutationExplainer, and other explainers (85). See the [docs](https://scalecast.readthedocs.io/en/latest/Forecaster/Forecaster.html#src.scalecast.Forecaster.Forecaster.save_feature_importance) The eli5 package appears to be deprecated.