Use endvelo states in compassUT

The MR !410 (merged) introduced the new Kalman beamline and endvelo states superseding the previous least means square fit beamline state. Using the endvelo states in the CompassUT algorithm caused an unexpected drop in tracking efficiency (these new states have actually better precision and improve also the momentum resolution), as discussed in #169 (closed). The issue is now understood and caused by the final momentum cut in the CompassUT algorithm.

This MR implements the endvelo states in the algorithm, change the final momentum cut (at the moment set it to zero) and removes the old lms fit velo states which are not used anymore in any part of the reconstruction.

Following the discussion during the presentation (https://indico.cern.ch/event/1033491/contributions/4348307/attachments/2238168/3794154/RTA_UT_pt_cut.pdf), the 2 possible choices are the "high efficiency" configuration with the Search Windows cuts P=1250,PT=200 and Final UT cuts P=1500,PT=250 which improves the physics performance and the "standard" configuration which reproduce the same physics performances keeping the throughput unchanged (Search Windows cuts P=1250,PT=275 and Final UT cuts P=1500,PT=400).

The final solution is to set the "standard" configuration inside the hlt1_pp_default while the "high efficiency" configuration is used inside a new sequence called hlt1_pp_high_efficiency.