MuonCombinedReco - Put back the layer intersections for Stau seeds, Fix segment sorting, SpeedUp extensions

Hi everybody,

this MR restores the MuonLayerIntersections for MuGirlStau, which were also killed in !48823 (merged). Instead of extrapolating the InDetCandidate through the MS, the TSOS of the MuidCo track are exploited:

1. The TrackStates are filtered for TSOS having trackParameters in the MuonSpectrometer
2. The muonLayerSurfaces are selected using the phi() angle of the calo-extension. This is quite the same to the ordinary approach
3. The TSOS which is closest to the surface plane is selected as starting point for the extrapolation. This reduces the extrapolation path from several meters to 10-20cm.

• As a performance present, I discovered that the algorithm can be made twice as fast by relaxing the requirements on updating the starting point for the extrapolations when going from one layer to the next of the MS. The physics impact has been tested on 20k data16 events and is found to be quite small (cf. All_quick_plots.pdf ). While the footprint reduces from

"MuonInDetToMuonSystemExtensionAlg": {
               "count": 20298,
               "cpuTime": 6941314.445365001,
               "malloc": 0,
               "vmem": 0,
               "wallTime": 7044486.0}

to

"MuonInDetToMuonSystemExtensionAlg": {
                "count": 20298,
                "cpuTime": 3494300.50838695,
                "malloc": 0,
                "vmem": 0,
                "wallTime": 3554827.0}

Tagging here @amete, @tcuhadar, @wleight

• The MuonCandidateTool associates the MuonSegments to the MuonCandidate. The association could be useful for later optimizations in thMuonInDetToSystemExtensionAlg and will replace the segment association in the MuonCreatorTool in a later MR
• The segments are now sorted following the same logic as in MuonSelectionTool. Tagging @mkbugge, @yoyamagu

Tagging: @goblirsc, @sabidi, @dcieri, @minions