Frustratingly, it is faster to run multiple queries with a single DISTINCT statement, cube them, and then join them together afterwards, rather than doing all the DISTINCT statements in one category
Tech debt:
Split out CLI verbose output from progress bar rendering, since the former should happen before a query is run, and the latter should happen after.
In files where this required adjusting imports, if they were not already doing module level versus class level imports, converted the file to the former.
Improved sorting for core testing data export in the hope that re-running it will be less noisy in git history.
cutover to cumulus__none as the blessed representation of null data from the source system.
Checklist
[x] Consider if documentation (like in docs/) needs to be updated
[x] Consider if tests should be added
[x] Run pylint if you're making changes beyond adding studies
[x] Update template repo if there are changes to study configuration
This PR makes the following changes:
cumulus__noneas the blessed representation of null data from the source system.Checklist
docs/) needs to be updated