Overall a lot of good information here. Thanks for doing this Joey. Below are my thoughts...
I am thinking the turbulance in the ports is slowing down the airflow at midrange and the more the rpms go up it start to streamline more with less swirl. Maybe with SVT cams the top will continue to gain over 6.5k more due to the unequal length headers.
Of course it is because the intake was designed to be used with secondaries, which is the whole purpose of the secondary runners. Maybe it will perform better with both the SVT cams and the secondaries.
The part that puzzles me is how the SVT UIM is the UIM was to small then how did it generate more torque over 6.5k than the taurus oval port when the top should be where the restriction should be more known.
It's designed around 1950's research. This isn't anything new. The design utilizes the "ramming effect". The intake is designed to provide peak performance through specific RPM ranges. It's using the wave pulses during RPM ranges to provide the most air at the given time it needs it at the cylinder. The SVT intake was "tuned" for a certain RPM range, which corresponded to the chosen runner lengths. By removing the secondaries you have taken out the impulses of air that it was designed to use at certain RPM ranges. The reason why it out performs the Taurus oval port intake at the top is because this is when the SVT intake is using all of the air it possible can at which point it was tuned to perform with both runners. The intake falls on it's face in the lower RPM's because your using both runners when it was tuned to use one per cylinder. I'm guessing if you run the numbers using some older formula's which Dr. Helmuth W. Engelman wrote about in his conference papers for ASME you would be able to conclude this with math as well. I would strongly suggest reading these research papers below to gain a better understanding of intake design before attempting to learn by trial and error. You may be able to save yourself a lot of time. Here is a list of worth while references...
Helmuth W. Engelman, The Tuned Intake Manifold: Supercharging without a Blower, University of Wisconsin – Madison, American Society of Mechanical Engineers – Oil and Gas Power Division Conference Paper, 53-OGP-4, April, 13 1953
Cambi, Enzo, Trigonometric Components of a Frequency Modulated Wave, Proceedings of I.R.E, Vol. 36, 1948, pp 42-49
Vorum, P.C, Short Pipe Manifold Design for Four-Stroke Engines, American Society of Mechanical Engineers- Diesel and Gas Engine Power Division Conference Paper, 76-WA/DGP-4, 1976
Vorum, P.C, Short Pipe Manifold Design for Four-Stroke Engines Part II, American Society of Mechanical Engineers- Diesel and Gas Engine Power Division Conference Paper, 80-DGP-6, 1980
I think for a true comparison you should run the SVT UIM , SVT LIM with secondaries and IMRC along with the SVT cams.
It REALLY is that BAD!
The biggest problem is that with the SVT UIM you have 2 runners per cylinder which flow at different velocities due to the unequal length of runners, then you are joining these before entering the combustion chamber causing a tremendous amount of turbulance where its not needed and at the same time reducing the velocity of the air where its needed the MOST!
Not a problem if it is used as it was intended. Again it was design around secondaries and specific RPM ranges for a 2.5L V6's volumetric efficiencies. If used with secondaries you wouldn't have flow issues through out the RPM range.
Just my .02. End Rant.
EDIT: Here is another reference book that is a little more up to date.
Read pages 309-315
ISBN # 0-7680-0495-0
Title: Introduction to Internal Combustion Engines
Author: Richard Stone
Publisher: Society of Automotive Engineers
400 Commonwealth Drive
Warrendale, PA 15096-0001
Copyright 1985, 1992, 1999