BF - you have brought up two differing cases; destructive testing and non-destructive testing. During the design phase, it is common to test to failure to see if the article meets expectations. If not, the designer will strengthen the design. If expectations are exceeded too much, the article will get "cut back" a bit. It may or may not be necessary to fully test serial production. If you do, for sure you will be using non-destructive tests. Proof is just such a non-destructive test. The gun is examined and measured prior to firing a known heavy load. After firing, it is re-examined for any damage or change in dimensions. Guns are designed in such a way and built of such materials that failure after proof testing is very, very unlikely, barring abusive use or conditions.

Your car design underwent a whole series of destructive tests before release to production and your individual car underwent a whole series of non-destructive tests as it progressed down the assembly line. The design of car building is such that it is not envisioned that the consumer needs to perform tests of his/her own beyond normal operation. Same with gun proof; the consumer does not need to reproof the gun on some schedule. Actually, only gunmakers are allowed to submit guns for proof (as I recall). They submit only guns that have had modifications or repairs that could possibly make the gun unsafe; for example, lengthening the chambers or honing/reaming out deep pits.

American guns are self-certified by the manufacturer, so we have no idea as to failure rates (although I'd be they were very low). European guns are tested at government proof houses, so there may be records as to failures. I'd still bet on low rates - makers can't afford to have big investments disappear in a flash of light and a puff of smoke.