"Streamlining" selections

[kreczko]

This issue refers tot he selection part of our analysis and to the merging of AS and NTP (issue #191)

In NTP we have:

• ttbar muon selection
• ttbar electron selection
• selection analyser (cut flows)

In AS we used to have

• BasicSelection
• ttbar muon selection
• ttbar electron selection
• various control region selections

My current understanding of what we need:

• base selection class that
  • inherits from EDFilter
  • monitors selection steps (cut flow) for both weighted and unweighted)
  • base functions similar to BasicSelection.h
  • adds the selection result as a boolean with prefix + selection step name to the event content
  • supports tagging mode
  • uses our final objects (jets, MET, etc) after corrections instead of applying them in the selection itself
• a python script that prints cut flow as (partially exists in DPS)
  • latex table
  • twiki layout
  • simple console output
• a script that can pick events passing a certain selection and storing the result in the format of {run:{lumi:[events, ...]}} aka JSON file. This one is useful for diffs of all kind (changes in selection, synchronisation exercise)
• one selection inheriting from base selection per signal- & control-region
  The last step should only contain minimal amount of code due to the inheritance.
  This would mean that most of the code in TopPairElectronPlusJetsSelectionFilter and TopPairMuonPlusJetsSelectionFilter.

This means that nothing would change in the way the selection are used in the python config (parameters will be passed down).

If everyone is generally happy with that I will start implementing the changes in parallel to the existing selections. This way we can validate old vs new with the described python scripts.