Most all actions are slightly different, so when cocked, how much, if any play is there?

If you had a solid lockup at the range, or for a range out the dining room window if possible aligned on a target of some sort, mark on a tree or whatever, say 50yards distance, you could feed (cycle) dummy rounds into the action and looking through the sights, see if there was a change in the sighting, between cyclings of the action - or between actual shots at a range - or merely test it by shooting.

Due to differences in rifles, there would be nor could be and guarantee or possibly and average of accuracy given - the accuracy of the average would be subject to the # of test rifles, would it not.