[BUG] Teach SeasonalMovingAverageModel to work with out-sample predictions

[Mr-Geekman]

🐛 Bug Report

Currently SeasonalMovingAverageModel and MovingAverageModel don't pass some tests written in test_inference. Try also to make sure that in-sample predictions work as expected.

Expected behavior

Tests are passed.

How To Reproduce

Run tests in test_inference.

Environment

No response

Additional context

No response

Checklist

• Bug appears at the latest library version

[GooseIt]

I've run the tests in tests/test_models/test_inference and after thorough examination have found that both MovingAverageWindow and SeasonalMovingAverageWindow pass all the tests.
However, upon examining these models more closely, I found two behaviours I believe to be bugs.
I link Colab notebook with their reproduction and description.

[Mr-Geekman]

It seems that we already completed this particular task, I'm sorry we forgot to close it.

About colab notebook. It seems like you are confusing classes. There are two separate classes: SeasonalMovingAverageModel and _SeasonalMovingAverageModel. The first is etna model that works with TSDataset, the second is private class that holds implementation details and works with pd.DataFrame.

[Mr-Geekman]

It seems like this task is already done during Inference track.