Multi-table should show foreign key transformers as None

[amontanez24]

Environment Details

Please indicate the following details about the environment in which you found the bug:

• SDV version: v1.0.0
• Python version: Any
• Operating System: Any

Error Description

In the multi-table case, all foreign keys are assigned None as a transformer. However this change is done in preprocess, and not in auto_assign_transformers. For this reason, if a user calls get_transformers after auto-assigning them, it won't show None for the foreign key. The proposed fix is to just move the code in _skip_foreign_key_transformations to inside of auto_assign_transformers. This will require some refactoring to loop over all tables and assign the transformers for foreign keys to None.

Steps to reproduce

from sdv.multi_table import HMASynthesizer
from sdv.datasets.demo import download_demo

data, metadata = download_demo('multi_table', 'world_v1')
synth = HMASynthesizer(metadata=metadata)
synth.auto_assign_transformers(data)
synth.get_transformers(table_name='City')

Observe the following output

{'Name': LabelEncoder(add_noise=True),
 'add_numerical': FloatFormatter(computer_representation='Int64'),
 'ID': None,
 'CountryCode': RegexGenerator(),
 'District': LabelEncoder(add_noise=True),
 'Population': FloatFormatter(computer_representation='Int64')}

synth.fit(data)
synth.get_transformers('City')

Notice the foreign key CountryCode now has None as a transformer.

{'Name': LabelEncoder(add_noise=True),
 'add_numerical': FloatFormatter(computer_representation='Int64'),
 'ID': None,
 'CountryCode': None,
 'District': LabelEncoder(add_noise=True),
 'Population': FloatFormatter(computer_representation='Int64')}