It's true that rear frame offset of even several mm is darn near impossible to notice while actually riding.
It's also darn unsightly to see the tire/rim way off the centerline of the brake caliper!


At the front fork however, even a 1mm offset at the tire's contact patch with the road will have the bike "pulling" (self-steering) to the opposite side that the contact patch is offset toward from the frame's centerline.


Threaded-rod method is ok to consider as a thought experiment, and I'm for free speech, but it is so time-consuming to unwind the nut(s) with each figurative "heave" on the chainstays to check on one's progress. It sounds like torture having to do it that way, not to mention the near-inevitable bending of only one chainstay, twice as far as one would want (ouch).


Someone mentioned using a straight piece of wood, ...I have never seen a straight piece of lumber.


A good "read" on dropout alignment is how "springy" that the QR lever action feels as one flips the lever closed. A springy feel that requires more than half of the lever's 180-degree arc to adequately clamp the axle is an almost-sure sign that the dropouts aren't parallel.
In the world of "flipper" bikes, this is generally the only check of dropout alignment I will bother with, unless wheel installation difficulty or inward/outward "rotation" of the rear derailer during QR securing is noted.


I have "upgraded" many five-speed bikes to six speeds where lesser-quality hubs are used simply by taking advantage of generous unused space between the freewheel and dropout. I might have to add only a 1mm washer to the driveside axle spacer stack to fit a standard 6s freewheel if 1) a modern chain is used and 2) there is no derailer mounting claw nut protruding into the available space. Campy and other higher-end hubs usually leave a lot less unused clearance so would need more spacer thickness added and corrective wheel dishing to go with that, and perhaps even a longer axle.