80x trigger update

[OlivierBondu]

fix https://github.com/cp3-llbb/HHAnalysis/issues/44

there is quite some new code, additional pairs of eyes and comments are very much welcome! (@BrieucF in particular as our trigger expert :wink: )

some notes:

• for dimuon paths: the legs have symmetric iso and id requirements (so far) but from the naming convention this could change, the recommended method is to check the proper HLT filters are required, as is done in the HWW trigger efficiencies using the muon POG studies
• for dielectron paths: no such thing about filters, at least no reference I could find in the HWW note (though it is present on the EGamma twiki)... however the iso and id requirements are symmetric for both legs according to the naming convention, so it is not needed and pt sorting of the hlt objects should be sufficient
• for different flavour dilepton paths: there is the PDG id in the hlt object, so this in addition to the naming convention is sufficient
• we should check if the slow turn-on of the various L1 seeds are properly taken into account in the T&P SF we use

This PR addresses only matching, and https://github.com/cp3-llbb/HHAnalysis/issues/103 is still opened

[OlivierBondu]

Additions in latest commits:

• adding missing filter for dimuon paths, spotted by @blinkseb
• as discussed offline, removing tkleg2 from efficiency computations (was set to 0 anyway, so this is just to clarify the code a bit, no change expected in the computed efficiencies or errors)
• adding the dielectron filters, as kindly provided by Arun's code

Incoming in a separate PR: update of all hardcoded stuff, then rebase of this PR onto it for validation that the code is correct

[blinkseb]

I have some concern about MC trigger path in triggers.xml. Currently, it'll match whatever double lepton triggers there's in the MC. For one event for example I have:

****************************************************************************
*    Row   * Instance *                                          hlt_paths *
****************************************************************************
*        1 *        0 * HLT_Mu17_TrkIsoVVL_Ele12_CaloIdL_TrackIdL_IsoVL_v4 *
*        1 *        1 * HLT_Mu23_TrkIsoVVL_Ele12_CaloIdL_TrackIdL_IsoVL_v4 *
*        1 *        2 *  HLT_Mu23_TrkIsoVVL_Ele8_CaloIdL_TrackIdL_IsoVL_v2 *
*        1 *        3 *               HLT_Mu30_Ele30_CaloIdL_GsfTrkIdVL_v4 *
*        1 *        4 *               HLT_Mu37_Ele27_CaloIdL_GsfTrkIdVL_v2 *
*        1 *        5 *     HLT_Mu8_Ele8_CaloIdM_TrackIdM_Mass8_PFHT250_v2 *
*        1 *        6 *     HLT_Mu8_Ele8_CaloIdM_TrackIdM_Mass8_PFHT300_v5 *
*        1 *        7 *  HLT_Mu8_TrkIsoVVL_Ele17_CaloIdL_TrackIdL_IsoVL_v4 *
*        1 *        8 *  HLT_Mu8_TrkIsoVVL_Ele23_CaloIdL_TrackIdL_IsoVL_v4 *
****************************************************************************

Notice there's both HLT_Mu8_TrkIsoVVL_Ele17_CaloIdL_TrackIdL_IsoVL (2015 trigger, now prescaled) and HLT_Mu8_TrkIsoVVL_Ele23_CaloIdL_TrackIdL_IsoVL (2016 trigger). If we decide to apply SFs instead of efficiencies, we should explicitly ask for the same triggers on both MC & data. Possibly triggers.xml can just be:

<?xml version="1.0" encoding="UTF-8"?>
<triggers>
    <!-- MC + Run 2016 -->
    <runs from="0" to="999999">
        <!-- Double mu -->
        <path>HLT_Mu17_TrkIsoVVL_(Tk)?Mu8_TrkIsoVVL_v.*</path>
        <!-- Double ele -->
        <path>HLT_Ele23_Ele12_CaloIdL_TrackIdL_IsoVL_DZ_v.*</path>
        <!-- Muon + electron -->
        <path>HLT_Mu23_TrkIsoVVL_Ele12_CaloIdL_TrackIdL_IsoVL_v.*</path>
        <path>HLT_Mu8_TrkIsoVVL_Ele23_CaloIdL_TrackIdL_IsoVL_v.*</path>
    </runs>
</triggers>

What do you think?

[blinkseb]

PR updated:

• Leg matching refactored
• Different flavor leg-matching was broken
• triggers.xml update removed
• Tools.h -> Tools.cc

[OlivierBondu]

Thanks !

side note: as the LHC apparently can reach 2e34 in luminosity I think it's unlikely we'll have a MC trigger menu in sync with data for the time being....