support fitted inverse transform in LocalStandardScaler

[jmoralez]

Description

The LocalStandardScaler didn't support inverting the fitted values, even if it didn't produce an error. This adds an invert_transform_fitted method to the BaseTargetTransform class since the logic is a bit different to restore the fitted values. Mainly, when predicting we know each serie has "h" values (the horizon), so computing the sizes is very easy, whereas when restoring the fitted values we have to compute the sizes, which can be expensive. Also when predicting we may only want a subset of ids, and on fitted we restore every serie.

Checklist:

• [x] This PR has a meaningful title and a clear description.
• [x] The tests pass.
• [x] All linting tasks pass.
• [x] The notebooks are clean.

[review-notebook-app[bot]]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB