DTF Double Peaks in MC for Downstream Sample

@shunan and I have noted these mysterious double peaks in DTF-constrained D and KS-mass variables in Block 1 MC (2024.W31.34).

This is for our analysis of B->D(KsPiPi)h. These double peaks are not seen in data, despite the tupling scripts and DaVinci version (v64r10) being essentially identical (lhcb-datapkg/AnalysisProductions!2086 (merged) and lhcb-datapkg/AnalysisProductions!2085 (merged) for data and MC, respectively).

The constrained B-mass variables look normal, while the following summarises the variables with the discrepancy:

D0_M (no constraint): fine
D0_M_KScon (only KS constraint): fine
D0_M_PVFit (only PV constraint): problematic
D0_M_KSAndPVcon (KS + PV constraint): seems fine, I don't know if it's the double peak too wide
KS_M (no constraint): fine
KS_M_Dcon: problematic
KS_M_PVFit: problematic
KS_M_DAndPVcon: problematic

E.g. signal MC on the left and data on the right:

Note that the signal MC has been truthmatched with the TRUEID's, MCMotherID's and MCGDMotherID's. These can be found in /eos/lhcb/user/m/mbovill/for_Laurent

Note also that this is seen in the Downstream, but not in the Long K-short sample.

See also the discussion on Mattermost - https://mattermost.web.cern.ch/lhcb/pl/npa87xx3r7n15jjdbbrqg8igic