break(datasets) Rename resplitter parameter and type to preprocessor

[adam-narozniak]

Issue

What is currently named as Resplitter= Callable[[DatasetDict], DatasetDict] and in FederatedDataset(..., resplittter: Union[Resplitter, ...]` is not the most accurate description what can happen.

Description

The resplitter can currently serve more as a preprocessor (it seems like a broader term to better capture the essence of the object).
After the dataset is downloaded, more operations that resplitting can happen. I'll list a few of them:

• removal of some data (not meeting certain criteria)
• transformation e.g., different date format
• creation of new features
  Currently, even though such operations can happen, they are captured as Resplitter (a callable that accomplishes the goal).

Proposal

Rename the resplitter parameter of FederatedDataset and Resplitter the callable to preprocessor and Preprocessor to better capture the essence of what can happen.
Rename the resplitter directory to preprocessor.

Also simplify the following names:

• MergeResplitter to Merger
• DivideResplitter to Divider

[jafermarq]

@adam-narozniak looks good to me, some comments below. I wonder if it would make sense to rename the resplitter directory to preprocessor.

[jafermarq]

Should we rename to TestPreprocessor?

[adam-narozniak]

I think for that I should divide the file so there would be one for merge_resplitter and one for preprocessor in general. I'll do after we confirm the directory renaming.

[adam-narozniak]

@adam-narozniak looks good to me, some comments below. I wonder if it would make sense to rename the resplitter directory to preprocessor.

We can do that, and I can move the preprocessor definition there, too. Maybe in the future, we will need to have more subdirs there, e.g., if we had more types of preprocessors.

[adam-narozniak]

Based on the discussion off the github I did the following:

• rename the resplitter dir to preprocessor
• moved back the preprocessor.py (renamed from resplitter.py)
• rename Resplitter to Preprocessor
• simplify the naming of *Resplitter to drop it from the names (so we have now Merger and Divider)
  (the description of the PR is updated accordingly)