I'm afraid I don't understand the concept (that's all right - there are lots of concepts I don't understand) and it seems to be the reverse of the conventional approach embodied in numerous applications, from jet engine inlets to instrumentation venturis.  The conventional thinking is that as the velocity drops the flow is more likely to become detached since there is less kinetic energy pushing the molecule forward and more pressure force pushing it back upstream.  The approximation for the "perfect" shape then turns out to be a straight tapered wall at about a 10-degree angle, terminating in a sharp edge.  The idea of an inlet is then to recover a significant portion of the pressure ahead of the inlet and then recover as much as allowed by the straight-wall diffuser and finally ignore the rest.  Typically 50% is recovered ahead of the inlet (that's about a 25% velocity reduction) and another 50% can be recovered in the diffuser.  That's a 75% total recovery, not too bad.  You can get better than that if the length available in unrestricted.  Take a close look at a subsonic jet engine inlet and note that the first part of the inlet is pointed inward and that's because at cruise flight the inlet is oversize and the airflow is deflected outward ahead of the duct.  The leading edge is rounded because at low airspeed the inlet is the then too small and the air is flowing inward to get to it, requiring the rounded shape to avoid separation.  Inside the duct it is more or less a constant diverging angle, rounded off just ahead of the fan blades.  The shape is modified because of the area of the fan hub "spinner".  Take a look at a carburetor venturi and you'll usually find the the diffuser increases in area and then is truncated, not bell-shaped.  The K&W picture of the "bad" diffuser shows the angle increasing to the point the air flow separates - just reduce this angle and it won't separate and that would, I would think, produce the best recover in the length available.  Then one might have to add baffles to distribute the low-velocity air uniformly across the core and that can be done with little energy loss.

Just some observations

Gary Casey