Doesn't having/using a case mouth 'stop' defeat the idea of measuring at various points on the neck - both rotationally and up/down - so as to identify the thinnest part of the neck?
New necks are tapered in thickness. While unturned, the thinnest part is the case mouth, and thickest is donut area. So if you're not setting a datum line, your measure will vary on that alone. Always measure one thing at a time.
Our brass is thickest at webs and taper thins all the way to case mouths by method of manufacture.
With this, thickness variance runs from webs to case mouths.
We measure necks for
both thickness and thickness variance.
You can turn away thickness variance at necks, but it's still running the remainder of case length.
Each time you size that variance, case runout grows, and before long you're making bananas for ammo.
This is the root cause of growing runout, so if you're intent on making straight ammo, you'll measure new necks for thickness variance, and cull out anything beyond 1/2thou as seen at any set datum.
And this is why a case stop is important for many of us (back to your concern).
As far as what matters for neck thickness itself; You determine this. It's your plan.
You probably want a set neck clearance, and consistent neck tension, including reducing sized or interfering donuts. This doesn't always mean you have to turn necks to get it. Depends on your brass, chamber, sizing, annealing plan.
I typically measure mid neck, with occasional measure on donut (just to know my thickness taper).
Believe me, if mid neck is thicker, donut is thicker as well. If there is variance at mid neck, donut and mouths vary as well. Both attributes are also offending shoulder to webs.
