Fixed a bug (by re-write from scratch) to make sure that the cumulative timeseries forecast is based off the same incident forecast that get submitted. Some bug meant these did not add up.
Added an incident-to-cumulative check when running the final submissions file.