User validation for update_transformers

[npatki]

Problem Description

When a user uses update_transformers, we should validate their input & throw appropriate warnings/errors.

Expected behavior

Perform the following 4 validations when using this function.

1. Users attempts to use update_transformers after using fit

Throw a warning, same as #466

ht = HyperTransformer()
ht.detect_initial_config(data)
ht.fit()

ht.update_transformers(column_name_to_transformer={
  'column_A': FrequencyEncoder()
})
Warning: For this change to take effect, please refit your data using 'fit' or 'fit_transform'.

2. Invalid column names

The column names must be in the config.

ht.update_transformers(column_name_to_transformer={
  'unknown_column_1': FrequencyEncoder(),
  'unknown_column_2': LabelEncoder()
})
Error: Invalid column names: ['unknown_column_1', 'uknown_column_2']. These columns do not exist in the config. Use 'set_config' to write and set your entire config at once.

3. Invalid transformers

The user must provide either a transformer object or None.

ht.update_transformers(column_name_to_transformer={
  'column_A': "FrequencyEncoder()",
  'column_B': 4.5
})
Error: Invalid transformers for columns: ['column_A', 'column_B']. Please assign an rdt transformer object to each column name.

4. Transformers are not compatible with the sdtypes

Check the sdtypes and throw a warning in case of incompatibility.

ht.update_transformers(column_name_to_transformer={
  'datetime_column': FrequencyEncoder(),
  'numerical_column': BinaryEncoder()
})
Warning: Some transformers you've assigned are not compatible with the sdtypes. Use 'update_sdtypes' to update: ['datetime_column', 'numerical_column']