I'm with the general flow of the others and echo that you shouldn't be trying this in a direct ativated style of control.
The full sized fighters typically fly with a neutral or negative stability margin these days and rely on the computers to control the surfaces to catch and limit the divergences that would occur if the flying was left up to the pilot alone. Without the various sensors and on board computer to make sense of it all you don't dare try to copy the full sized CG location and control system. Far better to keep the CG at what is normal for single ended control inputs (our typical RC style of flying) for forget about the aft CG closed loop arrangement.
However another option occurs to me. You could look at the idea of hinging the canards at the 23 to 24% point and then mass ballance the surfaces to a degree where they balance "just" in front of the hinge pivot and let the airflow of the moment "trial" the surfaces to whatever forms a low drag sort of positioning. A bit of grease in the hinge system would act as a viscous damping agent to avoid or at least limit any flutter but with the surfaces set up and balanced as indicated this should not occur I would think.
With such free flying surfaces when you're flying under high AOA on the main wing the canards would tend to at least operate as you've noticed by "flying" at a far lower angle of attack and coincidentally they would be directing air into the separated bubble when things get that far gone and may help with very high AoA flying by doing so.
Of course this is all guessing. I'd like to think that it's EDUCATED guessing but it's still guessing. A cheapie test model would be a nice way to try the idea out before commiting it to a fancy scale model. Something very cheap and dirty with a pusher glow prop engine and square fuselage that only looks Rafale'ish in a "stand way back and squint" sort of way...
