This run was done with fieldmaps that aren't exactly in the right place.  I went to the beginning of the QF and BD lattice reference arc and shifted the fieldmap forward by half the length in a striaght line (to put the fieldmap origin in approximately the right Z position).  So the fieldmap orientations are both lacking half the bending angle of their magnet, plus a small position error.

I'll redo this with exactly the survey geometry I specified later, although I also saved this configuration.

Compare orbits to Muon1 model tracking: only moved by ~1mm.  Compare tunes to Muon1 model tracking: very similar.