Some pawnless endgames, like mating with bishop and knight, or the queen vs. rook case you mention, just don't occur that much in practice, especially compared to pawn endgames or rook endgames. That means players are less likely to study them, and unfortunately for those players, in these endgames one often can't rely on general principles (rooks behind passed pawns or on the 7th rank, weak vs. strong bishops). I doubt masters would have any difficulties with a king + pawn vs. king endgame.
Also, let me quote Karsten Müller and Frank Lamprecht from their book Fundamental Chess Endings:
The queen always wins against a lone rook unless there is an immediate draw such as perpetual check or stalemate. The winning nature of the ending has been known for a long time and it was believed to be quite easy to force a win. However, the appearance of computer databases in 1978 caused the ending to be seen in a new light. While they confirmed that the queen should win from all normal starting positions, they also showed that it was quite difficult to win if the defender played precisely.