Yes, this is one of the problems with focusing too much on "accuracy" and thinking of recording as historical record primarily.
Actual performance practice historically has always been much looser, even according to those who've studied it in depth, than we assume. There's been far more room for improvisation than we tend to think. Heck, the cadenza in any concerto performance was not only an opportunity to show off one's skills playing the written music; it was an opportunity to improvise, compose on the fly! Right in the middle of the performance. With time, we've grown shy, and tend to play cadenzas that are known and written and chosen as "the best" (there's that "best" thing again.... <g>).
I think this tendency towards more precision has come with the loss of immersion in the culture and history and experience of the world of the music played; safer to just reproduce than produce, given that modern productions would probably mostly sound weak by comparison.
Take Artur Rubinstein, for example; he was a far "sloppier" player than, say, a Pollini, but listening to him, you feel you are connecting with someone who's still a part of the classical music world, lived it, didn't worship it as a museum piece to be studied to perfection.
Kind of like why I stopped listening to jazz, post-Wynton Marsalis. Nothing against him or his capabilities or approach, but the same thing has happened to jazz; it's become history, and died, and been turned into something to worship, instead of still being a living tradition that is extended and evolved.