As to mission-critical parts, there is some leeway in the definition of the term that may explain the confusion. "Mission-critical" means, necessary for the survival of the crew and passengers. The "mission" is, get to the desired location if you can, but abort safely if you can't. For passenger planes, this means that if one engine goes out, the plan can land on the other one (or two or three). If the ailerons go out, the pilot can achieve roll control with rudder and throttle. If the INS goes out, the GPS can get you to a visual on a landing strip. In each of these cases, there is a single failure and nobody dies. Hence, these systems are not "mission-critical" where the "mission" is transporting crew and passengers without loss of life. That the plane goes somewhere specific is a fringe benefit, so to speak.
IIRC, the O-ring disaster had to do with the poor design and decision-making regarding temperature limits of the inadequate O-rings. Once hot combustion products get past an O-ring, there will be a failure. On a system where those hot gases can impinge on a fuel tank, there will be a catastrophic failure. On the STS, that makes those O-rings mission-critical.
Interestingly, some have said that the choice of Thiokol is the ultimate cause of that disaster, for precisely the reasons I talked about in my original comment. The SRBs reportedly had to be able to be shipped over the road from Thiokol's plant. In order to do so (instead of building a plant near the launch site), the boosters had to be segmented. Segments -> O-rings. That's pure rumor, afaik.
Do I understand you right?
Do you have any sources? I am not an aerospace engineer, and I hate to pull that card after your long and thoughtful comment, but that seems like a pretty sensationalist claim, it'd be nice to have a little supporting evidence. Among other things, you'd figure if a spacecraft could be made as simple as a plane, it'd already have been done.
Also, does an airplane really have no single point of failure? I could swear everything does have at least one. I can't imagine how you could solve all of the single points of failure even on something as simple as a bicycle or a motorcycle. And with the problem with the O-ring failing- wasn't that the failsafe O-ring? I seem to remember hearing something about how there were two seals, and the first one was never supposed to go, but there was a second seal just in case. The first seal did go, but the operators just said 'hey, it's ok, we have a second seal'.