Test of truth matching needed

To be able to catch issues/bugs with truth matching like reported in Moore#736 (closed). Neutral truth matching should be considered too.

Example of a test that would have caught this would be running latest HLT2 over a sample to produce a DST and do DaVinci tupling step and run little script over it to print the list of matched TRUEIDs (frequencies) of certain final state particles and BKGCATs of composites.

Maybe needs the final test in integration tests? Essential is that there is a packing/unpacking step (latest bug was revealed by a change in the packing)