Technically, a setup like this is probably the best approach for multi-spectral scanning. Using a light source with a bank of multiple narrowband LEDs (best: integrating sphere), paired with a monochrome camera. Will take ages to scan and introduce challenging color science - I highly doubt that these guys got it right. At least on my display, there’s a persistent pink cast on all example images.
Let’s take a step back and look at different aspects of the whole “film digitalization”:
Digital Media can only approximate the old analog way: there’s no way current display technology will be technically be able to recreate the viewing experience of an audience sitting in a darkened cinema. For starters, the spectra of the light sources won’t match and you are viewing digital content normally not in a darkened room.
There’s a big difference between scanning color-reversal material vs. color negatives: as argued elsewhere, color-reversal film is seen by human observers whereas color negatives are “viewed” by print film or paper. The spectral sensitivities of human observers and print media do not match at all. Now: basically all color cameras are designed to get the colors right for human observers, closely matching the sensitivity curves of humans (XYZ). By design, they are not optimized for the more narrow-band-like responses of print film.
That is the reason why color-reversal material should be scanned into XYZ-space - which requires pairing a broadband illumination with any commercially available color camera. The color science of these camera is (should) be optimized for that usecase.
Negative material should be ideally scanned in such a way that the combined sensitivity of light source and camera matches the spectral sensitivity of the print material in question. The easiest way to accomplish this is technically is to use a multi-spectral camera with sufficiently many spectral bands. You can sort of “build up” the desired spectral response from the narrow sample bands. That is actually the color science in this case and it can be done. But it can be challenging. For example, the Viking Lander cameras had overlapping filter responses - something like that requires non-linear optimization to figure out a decent mapping.
Dynamic range matters (sometimes): a Bayer camera will always be less sensitive than a monochrome camera without any filters. That’s the reason why technically, RGB+W is superior to just RGB. The noise level of your RGB+W sensor will be lower than a comparable RGB-only sensor.
Also, with a multi-spectral camera, especially with heavily degraded material, you might be able to pick up a faint trace of image information in some part of the spectra which would simply be swamped out in the broad spectra of an RGB-sensor. A narrowband sampling enhances the signal variation which might still be left (that is: increasing the dynamic range of that specific channel).
However, it is quite challenging to get the color science right in case of heavily degraded material. If you are lucky, you have a “China girl” testimage somewhere in this footage. Than you have procedures at hand which can guide you towards the original colors of the footage.
But, in general, you end up just guessing colors (“ah, it looks nice!”) in post production. Whether the immense time and storage requirements of multi-spectral acquisition justify the effort in this case depends on the circumstances.
Aesthetics are important: I must confess that I gave up trying to reproduce the “look” of the original material for a long time. We’re living today in a different world with different visual aesthetics. For my viewers, shaky 18 fps footage doesn’t cut it. Also, too dark shadows are a no-go. Color nowadays is used to tell a story. Color variation due to illumination at noon vs. later in the evening (which the original analog color-reversal footage of course shows) might need to get ironed out - depending on what is shown. As I am producing my scans of old material for a current audience accustomed to Instagram and TikTok styles, I will stick to my broadband illumination + standard color camera approach. 
