Today I'll re-read the manual, more attention to details
"My guess is that the entire shutter switches R/G/B on successive frame
refreshes, but I don't understand why the complicated shutter circuitry is
necessary for that. I noticed that there are 2 lines (LCS_CSEL0/1) that
presumably select the color to display."
While I don't have a proof yet, I agree with you about successive frames. But I believe control commands of the shutter are serialized (only one or two wires, out of which one is already available - frame blanking) and then decoded in the (complex) shutter control board.
"Any color that's a combination of two or more prime colors would then
experience double (or triple for white) the exposure to the prime colors,
and would thus be brighter. It's also possible that the rendering is done
in four sweeps, like e.g. RGBW."
I think the complex colors would be diminished if seen through two or more layers of colors, more likely the same frame is successively exposed to different colors and the human eye makes the integration (I noticed that the working frequency of CRT is much higher than that for standard VGA, available at the back connector)
And for white would make no sense to have all fundamental colors ON; you just make all layers of color transparent.
I already tried Contrast pot a few weeks back, when I had all the video board out, same for Brightness and Screen, nothing wrong there.
Idea of checking voltages on the video board is worth considering, but it's painful to do it (should have checked that when i had it open) I believe it's good, but without proof... Ones again, those stripes synchronized to a clock signal... tells me the issue is before or at TLC34075.