The CAD drawing below shows my first design for the bracket. It bolts to the holes in the steering rack and moves the connection point for the tie rods upwards by 40mm.

A side effect of this is the the steering load acting on the rack is now trying to twist the rack internals more since the application point has moved further away from the rack centre line. The graph below shows the trade off between bump steer reduction and torque increase. "Normalised" data means that it has been divided by the standard data. So a value of 3 on the y axis means that the bump steer is 3 times worse than normal and a value of 2 on the x axis means that the torque is 2 times increased over normal. So the trade off for a 25mm wheel up-down move is either 3 times worse bump steer and no torque increase or no bump steer and 1.35 times the torque. Well, a 30% increase in torque will be OK I reckon given that the design load for a rack isn't exactly a precise number anyway.

A lot later I finally got round to designing a bracket. The photos below show the result and the stainless bolts. I'll be posting a drawing up sometime, but it's not complicated really. The bracket has two holes 50mm apart that lift the mounting point by 40mm. The bolts are all M12. The rack is M12 fine pitch, not standard coarse pitch. BTW, the bracket wasn't cheap! It cost me £50. That's because it was made from 316 Stainless Steel, the same material as knives and forks.

Here's some pictures of the bracket on the car.