improving 67 steering geometry

Yes, a custom centerlink will make a HUGE improvement. The pivot point for the inner tie rod should ideally be placed at the same height and spacing as the pivot point for the LCA (when wheels are straight ahead). The problem with this placement is that the travel of the steering box is so great that it will contact the oil pan of a V8. The solution is to greatly reduce the length of the steering arm on the spindle. I've done this with a bolt on steering arm on a Wilwood ProSpindle. The result is a tight turning radius like that of a modern car plus the reduced number of turns lock to lock required, all with no bump steer. See my write up on page 5 of my cardomain site.

The illustration is excellent in describing the relationships concisely for an SLA suspension.

Now, in the real world it's not necessary to have absolutely zero bump steer or true (100%) ackerman. For bump steer all you need is a close enough approximation of the LCA arc in the normal range of compression and extension to not generate a steering input that can be noticed or felt. For ackerman, all you need is enough to not squeal the tires in a hard turn.

I'm going by memory here, so don't quote me as gospel, but the slip angle of a typical P225/70 15" tire at 40 mph is in the neighborhood of 4 or 5 degrees (I'm sure you can get different numbers by doing a google search), and for an old bias ply of a comparable size it was around 7 degrees! That leaves a LOT of wiggle room. However, when you reduce the section height of the tire and increase the width such as on a P275/40-17, the slip angle drops way down to maybe 2 degrees or less. That makes bump steer very unforgiving, so you need to either fix it or don't drive with big low profile tires because you're dangerous.

As you can see from your illustration, yes the upper arm does play a role in moving line 1, and therefore the placement of the outer tie rod end to result in zero bump steer. In the Mustang world, at least with bolt on aftermarket parts, the placement of the upper BJ is very close to stock even if the arm is shorter (as it must be to result in a reasonable camber since the spindle is fixed), so that relationship will not change much at all, and the outer tie rod point is also fixed (on a stock spindle). Way overkill for ackerman, and at the cost of bump steer which is already bad because of the placement of the inner tie rod pickup point. In my design I move the outer tie rod pick up point longitudinally, while both inner and outer pick up points are moved outboard and the inner pivot point moves up as well. Therefore the arc of the tie rod very closely approximates the arc of the LCA from 1.5" of compression to 1" of extension which is the only region I was interested in. I have found in testing that it's not necessary, even with P275/40 front tires to have anywhere near the factory ackerman.

A lot of race cars have no ackerman because they keep the LCA and tie rods parallel and the inner pivot points the same because it will result in no bump steer without having to do any math or design work, but as your picture illustrates, it's not that hard to come up with a working geometry, and because there is some slip angle in the tires, it doesn't even have to be perfect, just "close enough". There is also another math model that will generate no bump steer, but unless you're studying to be a suspension design engineer, let's just leave it at that. The next generation of cars won't use steering linkage anyway.

In a stock Mustang suspension the IC is somewhere in imaginary land, as the imaginary lines don't intersect (at least not until you do the Shelby drop). I don't know what the engineers were thinking??? We were only a few years away from making it to the moon, so it's not like we didn't have the technology!