Implement forecast decomposition for SMA-based models

[alex-hse-repository]

🚀 Feature Request

Implement forecast decomposition for SeasonalMovingAverageModel model

Proposal

1. In methods forecast and predict add flag return_components: bool = False (see base classes of models)
2. Implement target components extraction logic for case return_components =True:

• Components of forecast are target lags depending on window and seasonality parameters, there should be window components at all
• Wights of the components are 1/window
• Names of the components are "target_component_lag_<lag_id>"
• Add target components to the result dataset via add_target_component method of the dataset

Test cases

• Test components sum up to target for different parameter sets (window, seasinality)
• Test that the component names are correct for different parameter sets (window, seasinality)

Additional context

Is blocked by Inference 2.1 milestone