Update LP SFs

[rkansal47]

• Wait for updated ratios from Oz
• Update subjet matching uncertainty
  • Keep weights from +1 and -1 prong reclustering
  • Store info about quark within [0.7, 0.8] and [0.8, 0.9]
  • Store info about double-matched (should already be done)
  • Store info about unmatched quarks
  • Use these weights in post-processing
• Distortion
  • Save LP ratios from signal in pickle
  • Do a run on all the signals with inference False and maxchunks 100 (100k selected events?) and no skimming just to get the LP ratios
    • Then accumulate and use those ratios for up/down distortion variation
  • Get signal / measured ratio in postprocessing
• Test LP SFs vs. Oz's code
  • Found kT subjet bug, now fixed by Oz (see pubtalk)
  • Subjets are matching
  • LP variables matching IF 1 GeV cut is removed (checked with Oz)
  • Nominal LP SFs matching
    • Had to ignore unmatched subjets like Oz (fixed 77% of jets matching)
    • Removed splittings below experimental resolution (100% matching!)
  • Uncertainties matching
    • Sys matching, except Oz code not using sys up/down pt extrap params (bug?)
• Fix top region normalization
  • top_has_bs gen selection fix
  • gen selection at the beginning of ttbar processor
  • Fix read_parquet bug due to empty parquet files
  • check selection and LP matches previous version
• Measure for resonant
• Re-do datacards and limits etc.