In such a case the crew would be dead whatever they did. Even if we assume there were enough suits for everyone to evacuate the Shuttle through the hatch (which AFAIK there weren't), and that people were able to evacuate before the fire caught them (which is unlikely), they would still be stuck in orbit with suits having only enough air for a few hours (the length of an ordinary EVA) and no way to get back to Earth.
Both of us are reasoning from a lack of access to shuttle schematics, but I'm always hesitant to say "never."
I was thinking of a fire that incapacitated the commander, was spreading, and the best available option was immediately venting all oxygen in the shuttle (probably with some crew loss of life), in order to potentially save enough systems to return home.
On the one hand, that's an extremely far-fetched scenario. On the other hand, NASA has experience with incinerating crews because they lack the means to save themselves.
> I was thinking of a fire that incapacitated the commander, was spreading, and the best available option was immediately venting all oxygen in the shuttle (probably with some crew loss of life), in order to potentially save enough systems to return home.
I think any such fire would make the Shuttle unable to return home safely anyway. And since it is, as you say, a far-fetched scenario, I'm not sure it would be worth trying to mitigate. Note that no such fire, or even any event remotely close to one, happened during any of the Shuttle flights (or for that matter any NASA mission after Apollo 1), which indicates that the probability of such an event was much lower than the probability of multiple other events that did happen to one or more Shuttles, including not just the two that were total loss of vehicle and crew but multiple "close call" events that could have but didn't.
More to the point for this discussion, since no such fire, or even any event remotely close to it, happened on any mission after Apollo 1, while there were incidents that gave cause for concern about the mental state of a crew member, it seems entirely reasonable to put more weight on mitigating the latter than on mitigating the former.
Indeed, which is why the actual policy that was adopted, as described in the article, is not what you describe. It is "the mission commander locks the hatch as soon as the Shuttle reaches orbit, and doesn't unlock it until deorbit prep". Whether to implement that at all on a particular mission appears to have been up to the mission commander, but if it was done, it was done that way, not the way you describe.
> the related rabbit hole is "How do you assure that commanders are maximally reliable?"
You put them through a much more rigorous screening process than payload specialists. The article explicitly draws this key distinction. Obviously no process is perfect, but the track record of the screening processes that are used for the things you mention is, AFAIK, extremely good. And, as the article notes, later on in the Shuttle program the process for screening payload specialists moved closer to the process that was already used for other crew members, and that had a positive effect.
Not in orbit. In orbit there is no scenario where you want to open that hatch. And the lock was only installed in orbit.