Years ago on a development engine that I calibrated, I received a rapid prototype layered aluminum intake manifold before the production-intent cast parts were available. This intake was pretty smooth. If the production casting was like a 220 grid surface roughness equivalent, the rapid proto intake was like an 800+ grit. In any case, the rapid proto intake made about 10 more horsepower peak than the production intake, and also saw a torque bump across the whole rpm range with no changes to EOI, cam positions, or spark. The performance gains were also backed up by an increase in the measured air flow and higher IMEP. This was on an NA high performance 4V V6 with a tuned induction system. Since I ran these numbers first hand, it made a pretty strong impression that the turbulent boundary layer along the walls of the plenum volume and runners is in fact significant. How much so probably depends heavily on the engine / manifold design itself.
I've been trying to find some real flow data for an extrude honed cast upper for years. There are a lot of anecdotal posts about extrude honed uppers "making a huge difference," but no hard data to show the flow benefit. I can see two potential benefits - runner smoothness and an increase in diameter. A light extrude hone which doesn't enlarge the diameter by much could reproduce the same benefits as the GT40 upper, if the fabricated intake's primary advantage is it's smooth wall tubing. I have a spare '94/'95 Cobra upper that I've been meaning to have extrude honed for some time, with the plan of getting it on a flow bench to document some real numbers. However, the cost has become prohibitive especially when you factor in shipping that big mass of aluminum back and forth. I might have to pick this back up and see if anyone in the Detroit area has extrude honing capability.
Beyond the smoothness factor, opening up the diameter of the runners will change the tuning of the intake in the right direction, albeit by a small amount. This would be a good thing for a performance engine since most of the Ford factory intakes are tuned for a low rpm torque peak with a displacement of 302 cubes. If you throw these intakes on a built motor with HCI or especially a stroker, it can push the torque down lower than you want it and it will result in the tuning going out of phase early at higher engine speeds for a lower peak power rpm than what your heads or cam are intended for. If you look at the intakes Saleen used on the S351s, they took a cobra intake and changed the plenum volume by a good amount to retune the Cobra intake for the 351 (you can see where they have a section welded into the middle of the plenum volume). The same benefit could be had on a built 302 or stroker motor. It would be great to find some CAD drawings of the stock intake manifolds to gather the exact dimensions and volumes. That info would make it possible to calculate the Helmholtz resonance rpms and determine how much volume to add to the plenum for a specific non-stock application and retune the peak torque and power points.