[SDV 2.0] Unify input/output for remaining synthesizer methods #3016

Environment Details

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

• SDV version:
• Python version:
• Operating System:

Error Description

With SDV 2.0, we want to unify the input/output for synthesizers across all modalities. A few methods still need to be updated to return a dictionary instead of a dataframe:

• SingleTableSynthesizer.sample_from_conditions
• SingleTableSynthesizer.sample_remaining_columns
• par.sample_sequential_columns

Additionally, these methods need to be updated to take the table_name parameter to be unified with multi table:

• SingleTableSynthesizer.get_transformers
• SingleTableSynthesizer.update_transformesr

Steps to reproduce

data, metadata = download_demo('single_table', 'fake_hotel_guests')

synthesizer = GaussianCopulaSynthesizer(metadata)
synthesizer.auto_assign_transformers()

room_type_transformer = OrderedLabelEncoder(
    order=['BASIC', 'DELUXE', 'SUITE'],
    add_noise=True
)

synthesizer.update_transformers(
  table_name='fake_hotel_guests',
  column_name_to_transformer={
    'room_type': room_type_transformer
})

synthesizer.fit(data)

condition = Condition(
    num_rows=10,
    column_values={'room_type': 'SUITE', 'has_rewards': True}
)

synthesizer.sample_from_conditions([condition])