Again Kai, you've come to fore with your schematics!

The concept of switching the current source via another opto hadn't occured to me - I was thinking along the lines of having an oscillator to control the pulse width. Nevertheless, Erik's suggestion has merit.
About the only negative I can think of is having optos side by side and back to front! There's always the possibility that in manufacture one gets put in the wrong way!

Kai - I note that you have used a transzorb as well as a MOV (metal oxide varistor). I've also done this on a few products - my thinking was that the transzorb is a little faster than the MOV - let the MOV take the worst of the hit and leave the reset to the transzorb, but reading up on MOVs suggests that MOVs are also quite fast in response. I've also found transzorbs by themselves fail the CE zap (Fast transients) test. I'd be interested in hearing your reasoning behind using MOVs(tm) and transzorbs(tm).

Obviously for low voltage circuits, the transzorb is more precise with the specified clamp voltage and is a better choice here and having the MOV means you can withstand a big hit and still have the transzorb survive.