Filter photons for PF

This MR tries to filter photons to be used in PF algorithm, to get the same event rate and BW as before changes reported in !4426 (merged). It is related to Rec!4345 (merged).

From a test using the QEE snippet described in https://gitlab.cern.ch/-/snippets/3438, this is what I get (for the whole QEE HLT2 lines):

before changes (aka default configuration)

BW = 1.161 GB/s
   LAZY_AND: Hlt2QEE_DiJetIncSVTag_pT25M40FullDecisionWithOutput                                                       #=10000   Sum=13          Eff=|(0.1300000 +- 0.0360321)%|
    LAZY_AND: Hlt2QEE_DiJetIncSVTag_pT25M40Full                                                                        #=10000   Sum=13          Eff=|(0.1300000 +- 0.0360321)%|
     DeterministicPrescaler/Hlt2QEE_DiJetIncSVTag_pT25M40Full_Prescaler                                                #=10000   Sum=10000       Eff=|( 100.0000 +- 0.00000 )%|
     VoidFilter/Default_Hlt1Filter                                                                                     #=10000   Sum=9608        Eff=|( 96.08000 +- 0.194071)%|
     VoidFilter/require_pvs                                                                                            #=9608    Sum=9489        Eff=|( 98.76145 +- 0.112833)%|
     ParticleRangeFilter/SimpleJets_689bc042                                                                           #=9489    Sum=548         Eff=|( 5.775108 +- 0.239471)%|
     TwoBodyCombiner/TwoBodyCombiner_d6f4405e                                                                          #=548     Sum=13          Eff=|( 2.372263 +- 0.650096)%|
     Monitor__ParticleRange/Monitor__Hlt2QEE_DiJetIncSVTag_pT25M40Full                                                 #=13      Sum=13          Eff=|( 100.0000 +- 0.00000 )%|
     Monitor__Global/GlobalMonitor__Hlt2QEE_DiJetIncSVTag_pT25M40Full                                                  #=13      Sum=13          Eff=|( 100.0000 +- 0.00000 )%|

after changes

BW = 0.928 GB/s
LAZY_AND: Hlt2QEE_DiJetIncSVTag_pT25M40FullDecisionWithOutput                                                       #=10000   Sum=6           Eff=|(0.06000000 +- 0.0244875)%|
    LAZY_AND: Hlt2QEE_DiJetIncSVTag_pT25M40Full                                                                        #=10000   Sum=6           Eff=|(0.06000000 +- 0.0244875)%|
     DeterministicPrescaler/Hlt2QEE_DiJetIncSVTag_pT25M40Full_Prescaler                                                #=10000   Sum=10000       Eff=|( 100.0000 +- 0.00000 )%|
     VoidFilter/Default_Hlt1Filter                                                                                     #=10000   Sum=9608        Eff=|( 96.08000 +- 0.194071)%|
     VoidFilter/require_pvs                                                                                            #=9608    Sum=9489        Eff=|( 98.76145 +- 0.112833)%|
     ParticleRangeFilter/SimpleJets_47bba9ce                                                                           #=9489    Sum=398         Eff=|( 4.194330 +- 0.205786)%|
     TwoBodyCombiner/TwoBodyCombiner_ef564b92                                                                          #=398     Sum=6           Eff=|( 1.507538 +- 0.610793)%|
     Monitor__ParticleRange/Monitor__Hlt2QEE_DiJetIncSVTag_pT25M40Full                                                 #=6       Sum=6           Eff=|( 100.0000 +- 0.00000 )%|
     Monitor__Global/GlobalMonitor__Hlt2QEE_DiJetIncSVTag_pT25M40Full                                                  #=6       Sum=6           Eff=|( 100.0000 +- 0.00000 )%|

I'm reporting results for the Hlt2QEE_DiJetIncSVTag line as it is the most BW consuming one ATM.

FYI @rangel @mveghel @rjhunter @isanders @ausachov @mramospe @mvesteri