I'll agree that losing the reference point is important, and that wear in the shaft opening is always eccentric, and typically located between 4 o'clock and 8 o'clock.
However, only fairly extreme wear will significantly affect rocker operation.
If the wear is .030" (the ID is now .822", very sloppy), the centerline error is only .015".
If the ID is simply enlarged concentrically (viz., no attempt to correct the alignment) to accept a bushing, the new fulcrum position will be too low by .015", raising the rocker on the shaft by .015".
If the rocker is adjusted normally, the long (valve-side) lever will be angled down by .42° (less than 1/2 of 1 degree) and the pushrod adjustment will be longer by .015" (shaft error) + (valve error ÷ 1.477), or .025". This is exactly 1/2 turn on the tappet adjuster.
In my opinion, although this permanently changes the rocker's geometry by changing the delta angle of both levers, this is not only harmless, but within the range of normal error due to wear and stacked tolerances.
Re-facing the pad, although obviously necessary, also changes the geometry by moving the long lever's angle upward, but again the change is almost undetectable. Taking off .010" to get a clean surface is about .3°. Unfortunately, it's added to any shaft center error.
The combination of wear patterns tends to identify rocker as suitable for rebuilding, or not. If the shaft hole ID measures more than +1/32" (top to bottom) and the pad surface is rough - use another rocker.
BTW: if anyone has 1954-* intake rockers that are too bad to re-use, I need some to cut up for research, happy to pay the postage and share the results?
