Saving thinned out clusters and drift circles to AOD

Adding thinning algorithm and flags to save pixel/SCT/TRT hits to the AOD, but only for tracks passing some configurable selections.

At the moment, all pixel/SCT clusters (TRT drift circles) and their MSOSs can be saved via flags.Tracking.writeExtendedSi_PRDInfo=True (flags.Tracking.writeExtendedTRT_PRDInfo=True). This MR adds a new thinning algorithm, ThinInDetClustersAlg, to PhysicsAnalysis/AnalysisCommon/ThinningUtils/ that thins out all hits not belonging to tracks passing some configurable selections. This algorithm is heavily inspired by the existing TrackParticleThinning tool in DerivationFrameworkInDet.

The algorithm is configured via preExec flags and is not added by default. Multiple track collections are supported, each with their own ExpressionParser selection string. While the hits are thinned, the tracks themselves are always retained. An example command for writing pixel and SCT clusters (and their associated MSOSs) to the AOD is shown below. Only hits belonging to InDetTrackParticles and InDetDisappearingTrackParticles with pT>10 GeV are retained.

Reco_tf.py --CA --steering doRAWtoALL \
    --preExec 'flags.Tracking.writeExtendedSi_PRDInfo=True; flags.Tracking.doTrackSegmentsDisappearing=True; flags.Tracking.thinPixelClustersOnTrack=True; flags.Tracking.thinSCTClustersOnTrack=True; flags.Tracking.thinInDetClustersTrackContainers=["InDetDisappearingTrackParticles","InDetTrackParticles"]; flags.Tracking.thinInDetClustersPixelMSOSContainers=["DisappearingPixelMSOSs","PixelMSOSs"]; flags.Tracking.thinInDetClustersSCTMSOSContainers=["DisappearingSCT_MSOSs","SCT_MSOSs"]; flags.Tracking.thinInDetClustersSelectionStrings =["InDetDisappearingTrackParticles.pt > (10* GeV )","InDetTrackParticles.pt > (10* GeV )"]' \
    --inputRDOFile [FILE] \
    --outputAODFile [FILE]

This work was presented in Inner Tracking CP meetings (March 11 and April 15), as well as the April 30 Reconstruction meeting. These slides discuss both the motivation (in the context of the SUSY pixel dE/dx analyses) and the implementation.

The envisioned use case is the dedicated custom reconstruction of a limited number of events. Taking the SUSY pixel dE/dx analysis as an example, the most interesting events can first be identified using only the information contained in the nominal AOD. Then these specific events can be re-reconstructed via the Event Picking Service to gain access to the pixel clusters.