Another response intended to change the way you're thinking about this (interesting that that is what all the posts are trying to do, while the OP is wondering if nobody has looked at the device his way before - I hope I'm keeping an open mind).
I deleted an earlier post comparing this to a control system because in retrospect I think it was unhelpful. There is a gain effect, but that controls the effective rating of the system, not the actual magnitude of the force that's being controlled. It's this distinction that makes me disagree with your assessment about the repeatability of explosive bolts. If you hang a 1 tonne weight from a (suitable) explosive bolt, the force released when you blow the bolt will be a quite repeatable 981N every time.
Whether that output force is constant or variable is a big issue. In a mousetrap, the output force is governed by the spring and is pretty much constant. At the other end of the scale, you have applications like parachute releases where the applied force may vary repeatedly from zero to several times the force that the mechanism is ultimately required to be able to release. In applications like that, you often end up with a design that gives you lots of gain, so that the required trigger force is never significant, coupled with a separate retention mechanism to stop the device triggering itself when there's no applied load (this can be anything from a spring or a pad of velcro to a shear pin). The effect of those is to make the required input force essentially load-independent - at which point, the sort of gain measure you were proposing loses a lot of its relevance.
If you were to come up with a simple "goodness" measure for trigger mechanisms, it would also need to include the distance the input force has to work over to take the unit from safe to tripped. The thing that determines whether a mousetrap is going to be a useful tool or a spectator sport is the amount of overlap on the locking bar - and hence how far you need to move the platform against the tiny frictional forces that are holding it up before it makes you jump. It might be that your goodness measure is activation energy - almost independent of rated load which is a secondary thing controlled by the number of stages in your device (many of these mechanisms are cascades of flipping levers and if you need to increase the force you're controlling, you just add more levers).
The "how much more I need to miniaturise it" piece really does have to be driven by the requirement, rather than by the achievable gain. Key factors will be:
[ul]
[li]How big does it need to be to sustain the service loading (which may be much greater than the force it needs to be able to release)[/li]
[li]How much space is available during the release sequence? (Unfolding levers can sweep a substantial volume)[/li]
[li]How much space is allowed during the times when the mechanism is not required to release? (In some applications - parachute releases are a good example here too - that might be a lot smaller)[/li]
[li]Are there assembly and inspection limitations? Make it too small and it may become unduly fiddly to put it together, and it might get too difficult to check that input latching mechanisms are properly engaged)[/li]
[/ul]
A.