If the sanity check is performed for multi-task challenges, the challenge attributes are grouped by the tasks. If the check is not performed, the data is not grouped.

Is there a reason why the sanity check is optional? Does the tool work correctly when missing data is not added during sanity check?

The check for cases appearing more than once per algorithm should definitely always be performed to avoid inconsistent results.