My experience testing red light versus sub-lumen white is that levels of white that allow me to perform tasks tend to be lower enough than levels of red suitable for the same tasks that it has less impact on my dark adaptation.
There are situations where that’s not the case, like viewing objects with very high contrast (e.g. black text on a white page) or instrument backlighting.
I think something that hasn’t been emphasized enough so far is luminous flux, as opposed to electrical or optical power. The lumen, and its derived units, lux and candela is calibrated to human visual sensitivity. 1 lumen of red light is much more radiant power than 1 lumen of green light. Much of the accumulated wisdom (not necessarily more robust scientific data) comes from decades of experience sticking filters on incandescent lights without dimming capabilities.
As an example, the Lee 106 primary red filter intended for entertainment lighting transmits 12.8% of the light from a 3200K tungsten-halogen incandescent source. I’m not actually sure if the transmission listed is luminous or radiant; if it’s radiant, the reduction in visible light is even greater.
Of course it’s no surprise to anybody that blocking most of the output of a light source mitigates its impact on dark adaptation and reduces its visibility to a faraway observer.
What would be interesting is a test involving extremely low-output light sources with very precise regulation in which subjects are given time to become dark-adapted, then gradually increase the brightness of their light source until they’re able to successfully perform some task, followed by a test of their dark adaptation.