enforced_latency should be according to Delphi time_value scheme, so it should be 5d back (Fri) rather than 4d back (Sat) from the Wednesday forecast due date.
Remove horizon -1 as it likely may trigger errors in the forecasters, and is no longer evaluated or visualized. (We may want to submit a degenerate distribution in its place reflecting what we think the latest value is, for purposes of FluSight runners checking "calibration". But that is not handled in this commit.)
Closes #37.
enforced_latency
should be according to Delphitime_value
scheme, so it should be 5d back (Fri) rather than 4d back (Sat) from the Wednesday forecast due date.Remove horizon -1 as it likely may trigger errors in the forecasters, and is no longer evaluated or visualized. (We may want to submit a degenerate distribution in its place reflecting what we think the latest value is, for purposes of FluSight runners checking "calibration". But that is not handled in this commit.)