Feature/trigger matching + double counting

[mafrahm]

This PR adds trigger object matching and cross cleaning between different data streams.
To achieve the trigger object matching, the trigger Selector has been reimplemented using a newly created Trigger collection from the config.

The strategy for the cross cleaning is as follows:

• Each trigger defines one data stream that it belongs to via a tag.
• We start with a mask that removes all events where triggers from this stream did not fire.
• In the config, we define the order of the data stream priorities. When a trigger from a higher
  priority data stream fires, the event is removed

Our data stream priority is data_mu>data_egamma>data_muoneg. Therefore events are kept as follows:

• a single/dimuon trigger fired --> event is kept in data_mu and removed in all other streams
• a single/dielectron trigger fired and no single/dimuon trigger fired --> events is kept in data_egamma and removed in all other streams
• a mixed (emu) trigger fired and single/dimuon and no single/dielectron trigger fired --> events are kept in data_muoneg and removed in all other streams
• no trigger fired --> events are removed in all streams

This means that we can (and need to) run over all data streams independent of the lepton channel when creating data/mc plots.

Currently, the trigger selection simply uses all triggers that are present in the cfg.x.triggers collection. To customize our dl selector, we can overwrite the trigger_config_func attribute via derive, which is a function that can modify the config during the init call of our selector.

TODO: we still need to implement these new selectors in the sl selector.