Brief discussion on section 5.1 (DE e.o.s.): Should maybe also show JLA alone and JLA + ZTF to show what we can get by just adding low-z data to current data.
Should maybe switched to Hounsell et al. prediction for the calculation (though that paper is not yet accepted it seems)
Also should stress that calibration matters (either shift contours or marginalize over calibration as nuisance parameter, thereby increasing the uncertainties).