statistical_inefficiency should raise an error when not given series

[VOD555]

The currunt statistical_inefficiency in preprocessing.subsampling will not do the subsampling when not given the series input and does not raise an error or any massage to show that it's not doing the job. This makes it misleading for one using it.

https://github.com/alchemistry/alchemlyb/blob/862343b7365114cef5aea82243715935f0af6e61/src/alchemlyb/preprocessing/subsampling.py#L530C10-L530C10

[xiki-tempula]

Sorry upon a closer look, this seems to be a feature which has been supported since god knows when.
When series is not provided, the statistical_inefficiency could provide a high level function that could be used to massage the data. For example, statistical_inefficiency(unsorted, None, sort=True) would sort the dataframe. I could however, issue an warning.