I was talking to our machinist about this, who one time worked on some parts of a satellite, and they were saying that the o-ring would have been fine. The temperature ratings for them like you said are well known and can withstand some pretty drastic fluctuations. The problem really was the design of the o-ring gland (more info in the Wikipedia article). If your O-ring isn’t sitting in the groove in the way it was designed to, it can easily become damaged and fail, and you and your rocket are fucked.
The Hubble faulty telescopic mirrors have a similar story. A testing engineer was brought out of retirement to ensure that the mirrors were properly ground. However, the manufacturing company, despite other signs that the mirrors were flawed, turned down the engineer’s request for some pretty simplistic tests that would have revealed the flaws. In fact, when the engineer (Rod Smith) would knock on the door to the grinding room, the techs would turn up the radio to pretend that they did not hear the knocking.