Analysis based on run 07290 only. For cuts I require exactly 5 space points in TKD but I have no other cuts included so far (not TOF nor number of clusters per space point). I make a linear least squares fit, allowing the linear least squares to make a track by track adjustment to the fit proportional to distance along the tracker (i.e. tilt the helix). I histogram those track-by-track tilt angles (theta_residuals.png). I note a mean of the fit of 2.8 mrad +/- 1 mrad (I have cut the tails in a statistically suspect way, however).
I also attach the residuals of the least squares fit; they look okay - I note tracker_1_station_3 has a slightly significant residual (about 1 RMS).
Compare with JHC,
https://indico.cern.ch/event/374187/session/6/contribution/29/attachments/745674/1022924/Mapping-CM42s.pdf
penultimate slide (32), which shows (roughly)
The physical bore:
x0, z0 = -3.9, 17650
x1, z1 = 2.3, 20300
gives tilt 2.3 mrad
The magnetic axis:
x0, z0 = -8.3, 17400
x1, z1 = -8.7, 20300
gives tilt -0.1 mrad (not significant I expect)
So the two numbers are consistent. Note that I need to check the sign...