Here's a dimensional comparison between the two bezels. As you can see everything is identical, including the color, with the exception of the display opening. However, the 5.6" LCD will fit inside the 8590 bezel, if I file it (or better mill it) to increase the horizontal opening by about 3mm, which is not a big deal, 1.5mm each side.
I played a bit more today with the video levels, I changed the resistor parallel with R301 from 390R (recommended by Newscope-OJr) to 220R and the separation between half brightness and full brightness is more evident. The camera kind of levels out the brightness but you can see there's difference between the grid and the trace.
Visual is even more obvious.
I don't know what mismatch you are talking about, but if you look at my previous pictures and the one that I attached now, the horizontal almost fills the active area of the LCD but vertically, there is some empty space top and bottom. Also, the grid is not square, which doesn't really matter since there doesn't really need to be a strict geometric correlation between vertical and horizontal.
More I refine my solution, more I feel like upgrading.
Last weekend I checked if I can hack the LCD to eliminate some colors and it works. Removing (or lifting) and grounding one side of the resistor networks that are series with the blue data bus from the LCD driver board to the LCD module, gave me a nice yellow display.
Edit: If you look at the picture "Figure 4 8590E Color Scheme #2" in the NewScope-0Jr_OM.pdf, it doesn't look any better than mine. Unless you use an LCD module with a resolution that matches the specs of the input signal, you'll always have interpolation.