Add preprocess method to synthesizers

[amontanez24]

Problem Description

As a user, it would be useful to transform my data without having to fit the model.

In order to allow users to separately transform and model data, the functionality should be split into separate methods. The first is the preprocess method.

Expected behavior

• Add preprocess method to BaseSynthesizer class.
• Parameters:
  • data (pandas.DataFrame): The data to transform.
• The method should do the following steps:
  1. Validates the data against the metadata (call synthesizer.validate)
  2. Transforms the data using constraints
  3. (if not already done) Creates HyperTransformer(s) + assign transformers
  4. (if not already fit) Fits HyperTransformers(s)
  5. Transforms data using the HyperTransformer(s)
  6. Return transformed data
• Note that for CTGAN, TVAE and CopulaGAN, booleans and categorical columns should not be transformed

Warnings

• If the model has already been fit, raise the following warning:
  Warning: This model has already been fit. To use the new preprocessed data, please refit the model using 'fit' or 'fit_processed_data'

Additional context

• We may want to add a parameter to the DataProcessor that let's us override defaults for sdtypes to account for the CTGAN, TVAE and CopulaGAN cases. This is similar to what we do in the current SDV version. We could also directly update the transformers to use in the HyperTransformer.
• If the HyperTransformer has not already been created, this method can call the assign_transformers method added in Add assign_transformers and get_transformers methods to synthesizers #1020