Best practice for saving Topo filtered candidates

At the BnoC WG we use as a prefilter for the inclusive HLT2 lines a topological MVA selection: we get from topological_b the require_topo_candidate function applying a cut on the MVA for the two-body and the three-body candidates. This practice can be dangerous because the information of what tracks made the filter pass is lost. Is there a "safe" practice to apply this prefilter? Maybe saving in the extra_outputs the tracks that made the filter pass is a good choice.

Another way of applying this prefilter can be implemented changing this HLT2 lines to the Sprucing. One could spruce on the output of the HLT2 Topo_2(3)-body line. Which is the best practice for triggering on these topo lines?

Any input from the experts is really appreciated, thank you in advance.

P.S.: this issue is related to merge request 1333