Append results (unless we are before the first timeseries start point)
Handle missing slots until the requested end (end_dt), but it might clash with the idea of reconstructors..
Do we want to allow 100% data loss before and after the end, breaks that a reconstructor reconstruct only between points. But, what if a slot has 25% data loss at the beginning and it is required to be reconstructed as well? This fails too..
Considerations: