Fit vertices about the track state reference point

[mgignac]

Kalman perigee states are defined about the beam spot (z = −1.1 mm), but the Billior vertex fit assumed the origin. As a result, target-constrained vertices were pulled to about −2.2 mm, closing the opening angle by about 1.6% and biasing the mass low (−0.9 to −3.5 MeV at 60–180 MeV in MC).

• HpsReconParticleDriver: new useTrackReferencePoint (default true) fits about the track state reference point. Set it to false for the old behaviour.
• BilliorVertexer.makeRk: the target constraint now uses the absolute vertex position.
• KalmanPatRecDriver: the conditions-DB beam position is only used when useBeamPositionConditions is set.
• ReconParticleDriver: optional setBeamSigmaZ.
• New 2021 MC steering file (14272) with the beam position fixed at (0, 0).

In MC this removes the target-constrained mass bias, apart from a ~0.5% track momentum scale. Unconstrained vertices are essentially unchanged.

[bloodyyugo]

Fixes the mass bias vs mass for TC vertex. This looks good to me.