I don't want to start a discussion between which polar alignment procedure is the best. I simply would like to understand the difference between phd2 and ekos while measuring polar alignment error. I performed the PA routine with ekos, getting a PA error < 37 arcseg:
But phd2 gives me a quite higher PA error (~7 arcmin) with its polar drift alignment tool:
Ekos shot frame do plate solving, rotate in RA axis and do another shot and then again. From this it can calculate where exactly lies your RA axis.
PHD2 on other hand just observe drift of selected star. In ideal PA it would stay static.
Flexing in scope will affect PA in ekos and in your case it pretty much average on target. Maybe next time try PHD2 routine then rotate RA for like 40deg and repeat.