HyperTransformer should sort columns after transform and reverse_transform

[npatki]

Problem Description

If the input columns don't match the order of the output columns, then users might think there is some mistake in the transform.

Expected behavior

If the original data has a column order A, B, C, ... then:

1. The transformed data should have the same column order with respect to the original column positions
   eg. A.value, B.value, B.is_null, C.value, C.is_null, ... (it is still the case that A is before B is before C...)
2. The reversed transformed data should have the same column order as the original

Additional context

Recreate this issue using the code below

import numpy as np
import pandas as pd

login_dates = ['2021-06-26', '2021-02-10', 'NAT', '2020-09-26', '2020-12-22']

df = pd.DataFrame(data={
    'last_login': [np.datetime64(i) for i in login_dates],
    'email_optin': [False, False, False, True, np.nan],
    'credit_card': ['VISA', 'VISA', 'AMEX', np.nan, 'DISCOVER'],
    'age': [29, 18, 21, 45, 32],
    'dollars_spent': [99.99, np.nan, 2.50, 25.00, 19.99]
})

from rdt import HyperTransformer

from rdt.transformers import datetime, boolean, categorical, numerical

field_transformers = {
    'last_login': datetime.DatetimeTransformer(null_column=True),
    'email_optin': boolean.BooleanTransformer(null_column=False, nan=0),
    'credit_card': categorical.CategoricalTransformer(fuzzy=True),
    # 'age': numerical.NumericalTransformer(null_column=False),
    'dollars_spent': numerical.NumericalTransformer(null_column=False, nan=0)
}

ht = HyperTransformer(field_transformers=field_transformers)

ht.fit_transform(df)

Notice how age.value is at the very end now.