I have thought about this myself (perhaps its in ASME?) but my expalnation is:
In the fire case the normal protection is deluge. If the deluge is in place then the vessel should get hotter, suffer no damage and see no higher pressure.
If the deluge fails then the vessel will get hotter - and the vessel is considered "lost". The purpose is now to limit escalation should the vessel burst. In this case you can save a buck and buy a smalle PSV because a higher pressure can be accepted when you have to discard the vessel anyway after the fire. No law against using 10% over pressure in the fire case as far as I know though.
Lizking don't think the probability of the event occuring has anything to do with it. If it happens then it should be safe - or you don't need it.
Koshan - don't think your argumentation makes much sense. If you need more capacity why not just buy a larger valve?
your explanation seems to make sense and I agree with your response to other suggestions. I would be interested to know if vessels really are scrapped when a PSV sized for fire relief lifts!