Tried another variation taking into account the area of the guide and valve stem on the width. To make the area follow the quadratic curve the width has to make a sharp jump essentially the opposite of the area curve. I put a fourth order curve fit through the points and made a port with ellipses of height and width at each station (blue line). The width has to come back to the valve seat diameter somewhat quickly. This port measured 95 CFM as printed.
Accounting for the guide and stem obstruction does not appear to make a big difference over the quadratic area port.
To eliminate the diverging walls of the quadratic port where it meets the seat I calculated a cubic area function with zero slope at the carb and seat.
[Linked Image from live.staticflickr.com]
This results in a wider port at the start of the guide (~3") with a shallower convergence angle of the width at the seat. This will be the next print.