Scott,
I guess I would give highest priority on getting the length along the battens the same, and then the length of luff and leech. If they are ok, a diagonal a bit off can only mean that the rise angle is not exactly right. That is not such a big deal. Weather the rise angle is 9.7°, 10.0° or 10.3°, is hardly noticeable (but it shows on the diagonals). This practice will ensure that the luffs of the four lower panels are straight, and that the panels fit together along the battens.
When bending that spline to get the right round, I place a nail at that 40% point. The idea is that the spline should not need to be forced into position at the aft end, but ‘find’ its position, thanks to the ‘40% nail’.
Accuracy:
Remember that this is not rocket science. We should be very satisfied with approaching 3-diget accuracy, and in most cases two digits is the best we can do. I mostly make my drawings with millimetre resolution, but am more than happy if I get close to the nearest centimetre during lofting and sewing.
The problem with getting the patterns perfect, only shows that practical construction of stuff is very different from producing accurate (computer) drawings. When I developed my ‘chain calculator’ method for finding the needed round and camber, others soon developed computer programs to do the job better. They surely produced results with impressing numbers of digits, but my guess is that the needed fudge-factors built into their algorithms did not produce more accurate camber than my primitive chain method does.
Finally, this shows the superiority of using paper patterns - much cheaper to do errors on paper than on the canvas ☺...
Arne