[FIX] web_kanban: wrong condition to reset the dataset index

This is related to rev. a218a9ed

The condition is good, but not in the right place: It should be done once all read_slice (all columns records) are fetched, not at each read_slice end