[PWGDQ] fix MFT momentum rescaling

[aferrero2707]

The MCH tracks are extrapolated to the first measured MFT point using TrackExtrap::extrapToVertex() instead of TrackExtrap::extrapToVertexWithoutBranson(), which cures large inaccuracies in the MCH momentum estimation at the MFT front for some tracks.
Those inaccuracies were contributing to high-mass tails in the di-muon invariant mass forward tracks with rescaled momentum.

Below is a comparison of the J/psi invariant mass before and after the fix, estimated from the same AO2Ds of LHC25i4 (OO MC simulation with injected J/psi and psi(2S)).

Before:

After:

[github-actions]

O2 linter results: ❌ 0 errors, ⚠️ 61 warnings, 🔕 12 disabled

[mcoquet642]

why not use the common tool for rescaling introduced in fwdtrackUtilities.h ?

[aferrero2707]

why not use the common tool for rescaling introduced in fwdtrackUtilities.h ?

Mostly to have the freedom of experimenting without affecting the common tools used in analysis.
For this specific case, if I understand correctly one could use refitGlobalMuonCov() for the momentum rescaling, right? However, the issue is in the determination of the MCH momentum to be used for the rescaling, so the common function would not help.

Once the fix is confirmed we can see how to properly propagate it to the common code. What do you think?

[mcoquet642]

why not use the common tool for rescaling introduced in fwdtrackUtilities.h ?

Mostly to have the freedom of experimenting without affecting the common tools used in analysis. For this specific case, if I understand correctly one could use refitGlobalMuonCov() for the momentum rescaling, right? However, the issue is in the determination of the MCH momentum to be used for the rescaling, so the common function would not help.

Once the fix is confirmed we can see how to properly propagate it to the common code. What do you think?

Indeed refitGlobalMuonCov does not fix the choice of the rescaling momentum (which indeed should be computed with Branson correction as you point out). But I think it would be good to be consistent with how the parameters are computed from the provided inputs. The one implemented in refitGlobalMuonCov has been extensively tested and is used by both EM and DQ (btw they also already use Branson correction to compute the rescaling of the momentum)