Move track shared hit info computation from Trk::TrackSummary to xAOD::TrackParticle creation [Solves ATLASRECTS-2807 needed for ATLASRECTS-7250]

This MR aims at removing the computation of the track shared/split hit information from the TrackSummaryHelper and the associated variables from the Trk::TrackSummary object into the TrackParticleCreator tool and the xAOD::TrackParticle variables. More details available in this short presentation.

In a nutshell, the changes in this MR:

• change the name of the unused fields in the Trk::TrackSummary to legacy_XXX (they cannot be fully removed because we want to still be able to read Run-1 data)
• remove the computation of the shared hits from the InDet::InDetTrackSummaryHelperTool + update the interfaces of the associated classes
• add the computation of the shared hits to the xAOD::TrackParticleCreatorTool
• adapt a few leftover tools retrieving the shared hit info from the Trk::TrackSummary to use the xAOD::TrackParticle instead
• adjust the configurations, both legacy and CA-based, to move the shared hit config from the TrackSummary to the TrackParticleCreatorTool
• introduce a systematic copy of the shared hit information from the original InDetTrackParticles to the GSFTrackParticles, which was previously only propagated for some algorithms
• introduce a systematic computation of the pixel expected hit + PID information for all the CombinedMuonTrackParticles, either from a proper computation in the TrackParticleCreatorTool for combined fit muons or from a copy from the InDetTrackParticles for the STACO algorithm (which do not have any associated Trk::Track available)

Outputs have been checked to be unchanged for trigger reconstruction, both in legacy and CA configs.

For offline reconstruction, expected changes are observed for

• the shared hit retrieved for GSFTrackParticles from the original InDetTrackParticles instead of being filled with dummy values
• the TRTTrackOccupancy and TRTdEdxUsedHits fields for the ExtrapolatedMuonTrackParticles and MSOnlyExtrapolatedMuonTrackParticles, which have their dummy values updated to some other dummy values
• the expected hit + PID information retrieved for CombinedMuonTrackParticles and CombinedStauTrackParticles (The pixeldEdx value already computed previously is unchanged).
• the shared hit information dropped for STACO muons in the CombinedMuonTrackParticles collection

Apologies in advance to the reviewers for the extent of the MR, but there's no real workaround around this.

FYI @goetz @sroe @christos @jojungge @stavrop