I would model the camera shrouds as pods, not fins. Give the pod a body tube that has the same frontal area as your camera shroud, give it a nose cone that's the length of any fairing you have at the front of the pod. Based on tests I've run before, I expect you will find the CP shift is much less dramatic, though drag may be comparable.I'm surprised the camera shrouds make that much difference in CP location and also in drag. If I move then to the bottom of the extension section which is aft of the CP and remove the nose weight, I can get to about the same 2.9 stability. One thing on this is that is that I have overridden the Cd of the shrouds to 0.25 because there lots of "holes" in them. I think this must have been close since my actual altitudes and sims were typically within 5-10%. The shrouds look like the attached picture. I no longer use an up camera, just redundant down cameras so I can redesign them with less "holes" and to be more aerodynamic. Note I am using cameras which are about a 1" cube.