We assume ISO timestrings are always UTC, and we deal with GPS time and TAI as continuous numbers only. Astronomers have the annoying habit :) of representing TAI as a YYYY-MM-DD HH:MM:SS style
timestring. It would be nice to be able to parse these in a situation like:

tt = spacepy.time.Ticktock(['2020-01-01T01:02'], dtype='TAI')

Right now this assumes the timestring is in UTC despite the dtype!

spacepy.time.Ticktock(['2020-01-01T01:02'], dtype='TAI').UTC
# dmarray([datetime.datetime(2020, 1, 1, 1, 2)], dtype=object)

We can of course do similar for GPS time. In both cases we could basicaly do ISO-style parsing with a fixed number of leapseconds (0 for TAI, 19 for GPS if I'm keeping head on straight).

This is going a little ways down the AstroPy route of separating timescale and format, but it seems a reasonable small enhancement.