Potentially fix bug in configuration of TrackResChecker

When comparing the fastest reconstruction as defined in https://gitlab.cern.ch/lhcb/Moore/-/blob/master/Hlt/RecoConf/python/RecoConf/hlt2_global_reco.py which is used in selection with the fastest reconstruction defined in https://gitlab.cern.ch/lhcb/Moore/-/blob/master/Hlt/RecoConf/python/RecoConf/standalone.py#L322, I noticed differences in the TrackResChecker. The TrackResChecker calculates resolution for Long and Downstream tracks. The difference between the two reconstructions is that in one case the tracks are merged with TrackContainerCopy in the other with TrackSelectionMerger. When applying this change where the containers are not merged for the standalone.py version but two instances of TrackResChecker are configured, I obtained identical results. I don't know exactly why.

In any case the current version is also just simpler, as tracks are not first merged to then inside the algorithm split them in different categories again.

Goes with Rec!2644 (merged)