Dave, re your previous question; whether its better to have small amount of bump toe-in or bump toe-out, I can only answer from my own experience.
I got the front static toe-in adjusted to 0.34deg overall (0.17deg per side). It turned out that the minimum toe change I could achieve from nominal to about 2.5" compression was about 0.12deg of toe-out on drivers side and 0.15 deg on the passenger side. So if a wheel goes over a bump, in theory it goes to near zero but doesn't change to a toe out situation. I'm not sure if this is the correct way to do it but don't have any bump steer that I can detect when I'm driving.
For your other point re changing rear castor; adding more castor the way you describe does change the static rear toe angle; i.e. increasing the castor (moving top of upright backwards) increases the rear toe-in. However, I don't believe it has much effect on the dynamic toe changes. The rear bump steer is strongly driven by the geometry of the lower (reversed) control arm and the lower radius arm. The vertical offset between those two arms on the rear upright means that the geometry of the upper control arm and radius arm will also affect bumps steer but to a lesser extent. That is why raising the upper control arm inboard pick up point significantly (as described by Derick and Howard) achieved only a partial reduction in the bump steer.
I looked into some of the commercially available suspension analyser programs, but unfortunately couldn't find one where I could include the GTD pick-up vertical offset on the upright design
I spent some time developing a software model of the rear suspension, and came to realize if you tried to look at (optimize) too many things simultaneously (bump steer, anti-squat, roll-centre control, camber compensation with roll) you were going to wind up with a compromise and have to decide what to prioritize. However the exercise gave me some useful insights. Raising the front pick-up point of the lower radius arm increases the bump steer toe-out and must be avoided. Lowering it improves the situation, but will lead to increased squat in the rear under acceleration. If you move the pick-up points outboard (away from the centre line of the chassis), then you can get a bit of anti-squat and reduce bump steer but you would have to fabricate new attachment points on the chassis. I did a bit of sensitivity analysis and was surprised how sensitive the best location was to tyre radius and ride height.
I believe that when GTD were designing the suspension, the were facing a bit of a dilemma. They would have known that owners would select a range of tires and ride heights so couldn't satisfy everyone, and rose joints don't solve all the problems.
On top of all that, I had to reverse engineer the geometry of the rear uprights so was always a bit suspicious of my computed results as there was some measurement error added.
So if anyone out there has an accurate dimensional drawing of the original style GTD rear upright I would be interested.
sorry for this over-length reply