Input validation & error handling in HyperTransformer

[npatki]

Expected behavior

We should check input and provide friendly tips, warnings and errors to guide the user into correct HyperTransformer usage.

Scenario 1: User calls fit or fit_transform directly without using the new auto_detect_config method (see #399)

>>> ht = HyperTransformer()
>>> ht.fit(data)
Error: No config detected. Set the config using `set_config` or pre-populate it automatically from your data
using `detect_initial_config` prior to fitting your data.

Scenario 2: User tries to transform a dataset that does not have the same column names or sdtypes as the fit data.

>>> ht.transform(different_data)
Error: The data you are trying to transform has different columns than the original data. Column names and their sdtypes
must be the same. Use the method 'get_config()' to see the expected values. 

Note: This should trigger any time that the set of columns is different. If the set is the same -- but the columns are just in a different order -- then we should proceed with the transformation.

Scenario 3: User tries to fit a dataset that does not have the same columns as the config

>>> ht.fit(different_data)
Error: The data you are trying to transform has different columns than the original data. Column names and their sdtypes
must be the same. Use the method 'get_config()' to see the expected values. 

Note: This should trigger any time that the set of columns is different. If the set is the same -- but the columns are just in a different order -- then we should proceed with the fit.