If it's going to cause paperwork problems, then don't go there. I completely agree.
But here's a thought experiment. Suppose you buy a multimeter, and after a few years in service, the voltage reference fails. The original part is unobtainable, so you replace it with a newer part that has a similar (or better) specification, and then send the repaired unit off to be calibrated.
It comes back reading bang-on, of course, with a certificate showing tiny percentage errors on all ranges. It's fixed, it's been tested, it works perfectly well, yet technically it's not quite the same as when it left the factory.
Is it OK to use? Valid scientific measurement of the unit at the cal lab would suggest 'yes', but there will always be a note in the equipment's log which states it was modified.
The question is, for what purpose is the meter now unsuitable, and why?
Could *any* testing restore adequate levels of confidence, for all purposes, in the meter?