Folks, we do not live in an absolute world. Anyone who has specified a vessel shell to be 1/2" thick and gone out and UT'd that shell and gotten a reported thickness of 0.500" please raise your hand.
As some of the previous responders have pointed out, life exists after the hydrotest. Real world issues include things like weld buildup to repair corroded areas, adding clips for various reasons, adding nozzles, etc etc. Many times the AI will not call for a hydrotest for these situations. Each situation is individual and needs to be assessed individually. Making absolute statements in a nonabsolute world is not realistic. Yes, I'd rather see things done perfectly. Yes, I've seen "R" stamps go on vessels before they were ever placed into service (no second hydro for the "R" stamp). Yes, I've seen "R" stamp nameplates inadverdently switched on major process vessels (in this case for rerates). Yes, believe it or not, I've seen a single vessel with two adjacent U stamp nameplates (long story there, and I don't agree with the logic).
It ain't a perfect world.
jt