If the camera chip is the only tilt contributor, agreed, the closer the better. Unfortunately I have no knowledge of how much, if any, is also introduced by the other image train components, such as spacers, focuser, OAG, filter wheel, backfocus adaptors and finally camera coupling adaptor. I will be attempting to evaluate these factors during my reassembly.
It's also the reason why I keep harping on about getting the basic OTA collimated before moving to the image train.
The trick I read about for the imaging train that has already worked well for me is the "laser mirror collimation" which involves taking the assembled image train, less OTA and camera, standing it vertically on a horizontal mirror and doing a laser return dot test. Simple and verifies alignment of everything except the camera chip, and most modern cameras have chip tilt adjustment anyway.
Might as well do these things anyway, because visibility outside here due to bushfire smoke is about 2km