I'm reading this late in the game, and I've been following the discussions about Oxygen for about 15 screen pages now. I work in the chemical industry and do research using pure oxygen on a regular basis. Oxygen is not flammable itself, but it supports combustion. Also, oxidation can generate heat. Gasoline and other substances have flammability limits (upper and lower) as well as autoignition temperatures. However, these carefully measured temperatures are for atmospheric pressures and concentrations. Increasing the O2 pressure or concentration can greatly lower the autoignition point. Increased surface area can also decrease the autoignition point, as a pipe of high-ignition-point liquid will autoignite if it leaks and soaks a high-surface-area insulation surround the pipeline. A cigarette will certainly ignite gasoline vapors, but as mentioned earlier it is dependent on the gas/air mixture being withing the flammable limits, and higher O2 pressures or concentrations will skew these limits. The friction created by gas flow through a small space at high velocity, as well as that created by a tight valve stem, can generate enough heat to ignite a minute amount of flammable material (organic or metallic) under conditions of high O2 pressure, or even under just increased O2 concentration. That is the origin of the "O2 Clean" principal. I have to clean equipment using rigorous procedures before introducing pure Oxygen under pressure, and as part of my training I have studied accident reports where you would never have believed there was a potential for a problem. Another ignition source (in the presence of fuel) not mentioned is static electricity, which is easier to generate that many would expect, it can be created simply by gas or liquid flow.
From the descriptions of the broken cylinder, including those of the burned inside walls, I would suspect that a minute amount of fuel was present inside the cylinder - either a touch of organic material or metallic powder. This could have been created by inadequate back-flow prevention when connecting to other equipment, slight impurities in the last fill, or deterioration in the valve seals. Hey, maybe even the (extruded out the side) valve-neck O-ring itself: how many folks wear clean-room gloves when installing a new O-ring? Skin oils are enough to start a fire in high-pressure O2, perhaps there was just enough to be below the auto-ignition point (for the condtions) until the shock of the falling cylinder. Once the flame starts, there would be an immediate pressure rise that would burst the tank. Burst disks on tanks are designed for simple over-pressure situations, not rapid pressure-rise situations resulting from combustion. When sizing reliefs for my reactor systems, any remote possibility of ignition requires huge (> 1 inch diameter) rupture disks to relieve the rapid pressure increase. Even then, we often are required to run the experiments inside a concrete and steel blast enclosure with walls 2 ft thick.
In summary, Oxygen does not burn but supports oxidation, and fire is simply rapid oxidation - think of rusting as a very slow fire (it does generate heat). There are many variables that can shift the flammability points any material (including increased surface area: dust), and increased O2 partial pressure (concentration and/or pressure) greatly reduce the ignition temperature, even to a point where it will auto-ignite. Myth-busters is simply entertainment and their testing is not all-inclusive; they are limited by budgets and entertainment values, and they are handicapped by a huge lack of real scientific knowledge, they rely more on anecdotal stories and a bit of light web-surfing for their test theories. It is interesting, though, to see how the anonymity of the Internet has made experts of so many regardless of education AND real-world experience. I work with Ph.D's that lack real world experience and make foolish decisions, but if we work together (example: I share my practical experiences and concerns and they calculate pressure-rise rates and maximums) then we can work safely. Too many folks figure that if they got away with something once it is safe every time, and this is not the case. Think of your buddies that overpressure their low-pressure tanks to 4000+ psi for cave diving or other extended range work!
By the way, it DOES sound like the cylinder was not secured and this allowed it to fall over when bumped or disturbed. A perfectly normal occurance that we all do, standing cylinders because "I know it's there and won't bump it". That's great until we do bump it. This sounds like a very unfortunate accident, but also a great learning experience for us all. I have great sympathy for the injured party and his family, a serious accident is heartbreaking, but a preventable accident is doubly disheartening. I find myself often donning safety equipment that I am SURE I don't need, just because it would be embarassing to have to explain how I got injured if something unexpected happened.
