Datetime reverse transform crashes if datetime_format is specified

[npatki]

Environment Details

• RDT version: 1.0.0 (dev branch)

Error Description

For both UnixTimestampEncoder and OptimizedTimestampEncoder: If you specify a datetime_format string, the transformers are not able to reverse transform the data.

Steps to reproduce

from rdt import get_demo
from rdt.transformers.datetime import UnixTimestampEncoder

# update demo data with a different datetime format
data2 = get_demo()
data2['last_login'] = ['Jun 26, 2021', 'Feb 02, 2021', np.nan, 'Sep 26, 2020', 'Dec 22, 2020']

# use a datetime transformer on this data: give it the explicit format string
ute = UnixTimestampEncoder(datetime_format='%b %d, %Y')
transformed = ute.fit_transform(data2, column='last_login')

# this step crashes
ute.reverse_transform(transformed)

ValueError: time data '1624665600000000000' does not match format '%b %d, %Y' (match)