That's really really cool. I thought it might be related to how glasses for colour-blindness work (eg. http://enchroma.com/technology/ - just the first one I found) but apparently they just use lenses with a specific transmissive spectrum to let through mostly lower frequency reds and higher frequency greens to improve differentiation.
I wonder if you could combine the two to give even more spectral resolution?
These filters were already optimized so the sum of their stimuli would match the tristimulus. (So you still perceive the same baseline as normal trichromats but with another asymmetry on top.) By removing a frequency range, you're just losing information and deviating from the standard perception, so I don't think it increase resolution, unless you were proto- or deutero-weak and needed the distinction boost anyway.
They do touch on expanding this by splitting all 3 photopigments, but that would be a much more challenging filter to make.
I was thinking that an active display could add more resolution by adding some time-dependent dimension to colors. Say, "high" reds/blues/greens would pulse quickly and "low" ones would pulse slowly... sort of a fourier transform aspect going on that could get you some scalar gradability rather than a straight up "left blue" vs "right blue" contrast.
I wonder if you could combine the two to give even more spectral resolution?