Hi Mike, If we can assume that the camring isn't chattering in the housing (there's a felt strip in the groove I think I can see? - that will help if there's metal-to metal wear), and that there is no axial float on the ring with the cover on (ie the mushroom still slips out of engagement even if you hold the camring firmly 'in' with your finger with the cover off) then it has to be a question of the parts and/or the geometry of the set up.
The amount of movement allowed by the large notch might be too great as you suggest, if someone has ground some metal off from one end or the other - hard to see from the pic. I wondered if some grinding had been done at the 'retarded' end from the pic, but probably just the light.
The largest undoctored 'clean' notch I can find on my bench, looking at a lot of K2F rings, is 11mm wide (using the short ears of a slide gauge to measure it quickly). Several are just over 10mm. Some have an angled cut, some are close to parallel-ended. If yours has been elongated, then even with a reasonably unworn mushroom and a decent V notch for it, things could slip out of engagement as you say, due to excess movement.
If all seems within limits (and the plunger is sliding smoothly in its bore, with the cable coming off the centre of it, not tending to push/pull it in a deviant way), then the only thing I can think of (apart from trying a different mushroom or camring), is to have a look at the stop or limit peg. The peg should be the 'sharp end' of an eccentric screw hidden under the pimple outside the cb housing. It should have been set in the right position for best internal mag timing when the thing was built/last rebuilt. However, if it is now at the extreme end of its useful movement (to provide maximum advance in this case), then it is just possible that this is what is causing the plunger to travel too far south and come out of the notch at full advance on the cable.
Unfortunately, it is hard to check the very best position for the pin, but as a rule of thumb the points should just be starting to open a few degrees after the magnetic flip points. If they open too early, you'll lose sparks at higher rpm (usually) but there is a bit of leeway, and you could try retarding it a wee bit by turning the eccentric nearer the centre of its movement to see if it makes any difference to the behaviour of the cable. It might, and there might be no obvious deterioration in sparking performance.
Every time we change the cb points, or put in a different camring, there is some small change in the exact firing point as wear patterns will be different. A new set of pattern points compared to a worn Lucas set can show up quite a difference, and I doubt most of us ever start playing with the limit stop eccentric to compensate because it is so hard to measure the difference unless something is seriously amiss.
Best of luck, and maybe someone will come up with a blindingly obvious better idea!
Cheers, Bill