After discussion with Robert Lupton on Slack the decision is:

• Add an "observation_reason" field to ObservationInfo (this is a string).
• By default, in absence of other information, set it to "science" when observation_type = science and "calibration" for every other case.
• For LSST test stands populate it from the TESTTYPE header. For test stands there is a finite set of allowed values so we have to decide what happens when something is encountered that is not in the set. Either we report it anyhow or we use "unknown".
• Finally, update Gen3 registry to add "observation_reason" as a string.
• To be consistent with observation_type I'll likely lower case them.

Simon Krughoff are you able to review this change? It's about 5 lines in 2 packages. Not a huge rush.

Tim Jenness the changes to these two packages seem fine to me, but with the discussion of using TESTTYPE to populate observation_reason made me expect changes to obs_lsst as well as the changes to obs_base and daf_butler. Is that happening elsewhere, or did I miss something?

That all happened in DM-26550 – this ticket split into two parts – the changes to the translator in one ticket and the changes to registry in the other. I needed to defer the registry changes until we could group it with other registry changes. I never updated the description of the ticket.

Fair enough. In that case, LGTM