Sort muon segments for the low-pT MVA working point
This MR implements a sorting of the muon segments for the low-pT MVA working point. This working point uses the chamber index of the first two segments, where "first two" was in rel. 21 simply defined as mu.muonSegment(0) and mu.muonSegment(1). This would give generally the innermost segments, which is no longer the case in rel. 22. The order of the muon segments in rel. 22 appears to be "random", and thus some sorting is needed to reproduce the rel. 21 results, or in general for a notion such as "the first two segments" to be meaningful.
What I did for now was to match the ordering scheme to what I empirically deduced from a rel. 21 test sample. This is in the end not a strict innermost-to-outermost ordering, but when applying the rel. 21 training of the BDT, I guess we should stick to the rel. 21 ordering. However, for a rel. 22 retuning of the low-pT MVA working point, one could certainly consider changing this order. This can be very easily done with the current implementation.
I considered a few different ways to implement the sorting technically. I think what I ended up with is both readable and efficient, in addition to having the possibility to easily change the order if desired/needed. But I am open to suggestions if there is room for improvement.
Some slides that demonstrate the need for sorting are found here.