Move main Jet/MET reconstruction to derivation step, switch to FlowElement as default
This is a major change to the Jet/MET workflow, with several components:
- FlowElements are now the only PFOs stored in AODs by default. They're now called
JetETMissChargedParticleFlowObjectsandJetETMissNeutralParticleFlowObjects, which is the name previously used byxAOD::PFOcollections. The specialized "EOverP" collections are unchanged. - Isolation calculations have been updated to use FlowElements as inputs.
- Jets and MET (and flavor tagging) objects are no longer stored in standard AODs. The reconstruction of them is still scheduled for monitoring purposes. Configurable flags have been added to restore them to the AOD content if the user chooses to.
- The standard EMPFlow jet collections now use FlowElements as inputs instead of
xAOD::PFO. - MET association building now uses the FlowElement links for electrons, photon, muons, and taus by default instead of post-hoc DeltaR matching.
- The derivation framework now supports all standard jet collections and MET.
- PHYS, PHYSLITE, and PHYSVAL now schedule the reconstruction of the required jet collections and MET objects instead of copying them from the AOD.
- None of this affects any trigger workflow - this is purely for the offline reco chain.
Given the nature of these changes, derivations in a release from after this MR goes in using the standard configuration will only work on AODs made after this MR goes in. For example, derivations in 22.0.40 will generally not work on AODs made with 22.0.37. Having a clean break between the two is the main motivation for doing this in one omnibus MR.
I've tested this workflow locally on data from RAW through to DAOD_PHYS and each step has the correct content as far as I can tell. I'll run more local tests with MC and PHYSLITE/PHYSVAL in parallel with the CI.
Tagging:
@sawyer, @cdelitzs, @mswiatlo for Jet/MET
@duperrin, @sanmay, @cschiavi, @vdao for FTag
@mhodgkin, @goetz, @elmsheus, @emoyse for central Software/Reconstruction