Integration of BSM tuning of PrLongLivedTracking

A few points to consider the integration of the BSM tuning of PrLongLivedTracking:

• Make it possible to distinguish tracks from the nominal and the BSM tuning of PrLLT. Can be achieved by adding a History field to the PrDownstreamTracks container, similar to PrLongTracks here, and then propagate this through the Kalman Filter. LHCb!5182 (merged)
• Check how to best combine both containers. Prototype is here
• Tune the combination of both algorithms to not (significantly) increase the ghost rate, potentially adapt the ghost probability cuts.
  • Either tune the ghost prob cuts for tracks from both tunings. This has the disadvantage of potentially breaking consistency for all tracks wrt to the beginning of 2025.
  • Or apply different cuts for the two tunings, which would create a non-smooth ghost prob distribution. @mveghel @samarian