MuonCombinedReco - Refactor algorithm dependencies

This MR is a keen attempt to reduce the CPU footprint of the combined muon reconstruction. As outlined in ATLASMCP-107, the combined reconstruction schedules independently, segmnet-tagging, calo-tagging, STACO, MuidCo and MuGirl of each other. In the vast majority of the cases, preference is given to tracks reconstructed by MuidCo.

Given that MuGirl and SegmentTagging are quite CPU exhaustive, the idea is to split the creation of the MuonCandidates into two steps

1. Select the ID tracks and augment them with a calo extension. Pipe this collection to MuonCaloTag and MuonCombined
2. Evaluate which inner detector candidates are succesfully combined and exclude them from the candidate list. Equip the remaining ones with the muonSystemExtension used by MuGirl to find the combined tracks. Pipe the new collection to the SegmentTagging & MuGirl

To perform step two, the MuonInDetToMuonSystemExensionAlg is added. The constraint to perform the extrapolation into sectors only, which was added in !48130 (merged), is moved from the MuonInDetCanidateAlg to the MuonInDetToMuonSystemExensionAlg. However, without removing the hits already used up by MuidCo. Tagging @iliadis, to implement these in a following MR.

To ilustrate the gain in CPU footprint, a reconstruction test is performed on data18_13TeV.00357750.physics_Main.daq.RAW._lb0105._SFO-4._0002.data using the bare 2021-12-03 nightly

15:44:58 PMonSD          n    cpu    max@evt    vmem    max@evt  malloc    max@evt  component
15:44:58 PMonSD [evt]   99    302    700@66        0      0@0       357    859@86   MuonCombinedInDetCandidateAlg
15:44:58 PMonSD [evt]   99    140   1020@66        0      0@0       127    901@66   MuonInsideOutRecoAlg
15:44:58 PMonSD [evt]   99     88    660@66        0      0@0         1      6@40   MuonSegmentTagAlg
15:44:58 PMonSD [evt]   99    100   1430@69        0      0@0        84    499@69   MuonCombinedAlg

15:44:58 PMonSD [evt]   99    196    990@93        0      0@0       221   1082@93   MuonCombinedInDetCandidateAlg_LRT
15:44:58 PMonSD [evt]   99     43    390@93        0      0@0        58    474@1    MuGirlAlg_LRT
15:44:58 PMonSD [evt]   99     48    240@27        0      0@0         0      2@72   MuonSegmentTagAlg_LRT
15:44:58 PMonSD [evt]   99     32    600@32        0      0@0        11    211@32   MuonCombinedAlg_LRT

and the modifications introduced in this MR:

13:58:55 PMonSD [evt]   99     21     60@86        0      0@0       290    728@86   MuonCombinedInDetCandidateAlg
13:58:55 PMonSD [evt]   99    275    650@57        0      0@0        68    185@8    MuonInDetToMuonSystemExtensionAlg
13:58:55 PMonSD [evt]   99     72    670@66        0      0@0        62    554@66   MuonInsideOutRecoAlg
13:58:55 PMonSD [evt]   99     80    610@66        0      0@0         0      3@66   MuonSegmentTagAlg
13:58:55 PMonSD [evt]   99    100   1440@69        0      0@0        85    500@69   MuonCombinedAlg

13:58:55 PMonSD [evt]   99     12     40@27        0      0@0       177    856@93   MuonCombinedInDetCandidateAlg_LRT
13:58:55 PMonSD [evt]   99    183    940@93        0      0@0        48    234@93   MuonInDetToMuonSystemExtensionAlg_LRT
13:58:55 PMonSD [evt]   99     42    390@93        0      0@0        57    474@1    MuGirlAlg_LRT
13:58:55 PMonSD [evt]   99     46    230@27        0      0@0         0      2@72   MuonSegmentTagAlg_LRT
13:58:55 PMonSD [evt]   99     31    590@32        0      0@0        12    211@32   MuonCombinedAlg_LRT

The MuGirl alg takes about 50% less CPU time and SegmentTagging about 10%. Relatively small gains are achieved for the LRT chain and for the construction of the full InDetCandidates. This will be further optimized by @iliadis.

It is super crucial to note that this approach will defeat basic assumptions in the strategy to measure the reconstruction efficiency on Jpsi candidates. Segment tag probes will only include the muons rejected by the bulk of the combined reconstruction. Tagging @dcieri, @gabarone, @mvanadia for information.

Also tagging @wleight, @rosati, @stavrop for a review-pending-expert

Addresses as well ATLASRECTS-6726, ATLASRECTS-6734