breaking: remove deprecated behavior

[jmoralez]

Removes the following behavior, which has been deprecated for about a year:

• Providing the id as the dataframe index, it has to be a column. Also the outputs from the methods will now always have the id as a column.
• Providing sort_df to fit and cross_validation, we always check if it's sorted.
• Providing num_workers_loader to the models constructor, those should now be provided through dataloader_kwargs.

Also updates the documentation to remove usages of this behavior, docs that use datasetsforecast.losses and uses utilsforecast.plotting.plot_series instead of StatsForecast.plot

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[elephaint]

Mainly non-blocking comments, but the following notebooks in tutorials still have os.environ['NIXTLA_ID_AS_COL'] in it:

• 02_cross_validation
• 10_distributed
• 12_using_mlflow
• 19_large_datasets
• 20_conformal_prediction

[jmoralez]

I removed the uses of datasetsforecast.losses, the ones that set the env variable wouldn't be broken but it can be confusing, so I removed them in 051ec9b. We still have to update all notebooks which have a reset_index in them but I wanted to finish this within my lifetime so I didn't do it here but I'll update this PR when I get time to run those as well.

[elephaint]

Thanks!