Basically, I'd say this isn't a rule.
Or maybe it's a misphrasing of a rule.
You generally see the last bar shortened to match the pickup bar when there's a repeat that goes right back to the start of the pickup bar, in order to preserve overall bar lengths as consistent within the section. This then implies that the subsequent section also has a pickup (the remainder of the bar with the repeat marking in the middle of it).
And notationally, actually, that's often done by dropping a start repeat marker on the first full bar after the anacrusis, so the anacrusis for the second time through is actually incorporated into the last bar of the repeated section. This enables the anacrusis to be different the second time, but more importantly means you don't have to stick a repeat sign in the middle of a bar, which can be confusing to sight-readers.
With regards to the end of the piece, I've seen a few pieces where the piece ends on a partial bar, but I've seen a lot more pieces where it ends on a full bar, thus leaving the entire piece with a non-integer number of complete bars.
My experience is primarily with baroque, renaissance and English folk music, so this might be different elsewhere, but the pattern I generally see is:
(pickup bar):(lots of full bars):(finish with a full bar, sometimes a longer bar)
In an era where throwing in bars of different lengths from time to time was perfectly normal (and indeed a lot of music was being written without bar lines anyway, making that kind of thing much more tempting), not 'matching' your pickup opening in the last bar would've been entirely unremarkable. If you look at scores for various fantasies, ayres, various dances and so forth you'll often find the ones that do start with a pickup still finish with a full bar for the final chord (at least in the bass, sometimes in every part).
So I'd only expect to see anacrusis length matching when repeating.