Is it reasonable for a heavy door to "open by itself" (ie from differences in air pressure) if it had already been slightly ajar? If you consider a basically uniform massive door (say, 300 N) where there is some coefficient $\mu_{s,k}$ of static  and kinetic friction between the thing on the inside of the door, and where the door is 1 meter long (1 meter from pivot point to end of door), is it reasonable that differences in air pressures inside and outside of the room could cause the door to open slightly, seemingly by itself, if it started where the metal thing was not snugly in the hole? 
 A: There are two ways a door can open slightly by itself. One is as you mentioned air flow and air pressure, and the other is if the hinges are not exactly vertical and/or aligned properly. Especially since the wooden frames move with time and the hinge screws tend to loosen doors may be come more and more likely to not have their most stable position as closed.
As far as the air pressure goes, think about it this way. While the door is closed, pressure builds until it is relieved by the door opening, or the air flows out from somewhere else (windows cracks). So the total pressure than can be reached depends on the configuration.
A: Yes, provided the air pressure change is rapid enough (simple air pressure changes due to weather won't do). It happens if the rooms has closed windows, two doors, and the room is fairly small: you slam one of them, and the other closes (or opens). It happens all the time in my bathroom. 
An explosion would do as well.
If the door connects a hall and a room, the hall is narrow and the door is light, someone running in the hall could produce enough air pressure to open the door, too.
I find it unlikely that it could be triggered by temperature differences between two rooms (provided the temperature difference is not increased abruptly), because even though temperatures were different air pressure should be roughly the same for the aforementioned reasons. 
