It seems our problem is now solved thanks to a related observation in [Import C-100 tokens] and the solution offered by its author (koenr - thank you!). In our case, we wanted to import the seeds from Google Authenticator files and that format requires a conversion. Using the code provided by koenr:
import base64
import binascii
...
key_bin = base64.b32decode(seed)
key_hex = binascii.hexlify(key_bin).decode("utf-8")
where seed is the first line of a .google_authenticator file and key_hex needs to go into the seed column of the csv file.