The problem with lighting is not the fact that energy levels of constant intensity increase with the frequency of light waves.
Our eyes adjust the iris to balance the energy level of what we see by adjusting the effective aperture size much like a camera.
We have poor acuity in the dark because the blue-green light is what affects the iris and not red , so we do not get night blindness in red light but can get night blindness from bluish HID headlights. Our eyes have a slow adaptation in Scotopic ranges of about 30 minutes but in seconds for bright Photopic ranges.
We know that if we stare at any bright light for a long time, we see spots in the dark from the memory pigment effects. If we stare at a very bright 10cd red LED it slowly changes colour towards yellow from adaptation. If we stare at gas tube baylights the memory effect is pink spots. If we stare at the sun which is >1kW/m^2 or 100k lux (cd/m²) we know it can burn our retina after a couple seconds which is only broad spectrum power level of 1 mW/mm² yet if we view reflected sunlight with 10% average reflectance levels of 10k lux , we might be inclined to wear sunglasses, but oftendont.
However if surrounded by snow, with 90% reflectance we must wear glasses with slits like the Inuit once did or sunglasses but in 1k lux we know it is very comfortable like the fluorescent ceiling lights recommend.
Bright direct sunlight >100,000 Lux
Photopic normal daylight 100 to 10,000 Lux or cd/m²
Mesotopic Twilight 0.001 to 100 Lux or cd/m²
Scotopic Night vision 10^{-6 to −3.5} cd/m²
We see that our eyes have a dynamic range like our ears with 11 decade range, but we know that we can't hear a pin drop when standing behind a jet engine.
Our typical useful dynamic range is 2 to 3 decades which requires great attention to go beyond this with A-B comparisons.
Our vision is the same with 2~3 decade range and in TV they call each unit of Luminance = 1 IRE going from 0 to 100 IRE . We often never can see between 0 and 1 IRE due to the glare on the TV glass from ambient light or poor adjustment or capability of the black level is in this range. Thus our perception in this narrow range of our wide dynamic range is fairly linear. Most TV's and monitors are capable of 250 Lumens/m² but if up close, our eye strain can cause fatigue or headaches so reducing it 50% or to the minimum acceptable level permits us to use a monitor all day if needed and not suffer.
I would turn it around and say , give us a spec to meet and we'll meet it.
But if they balk from ignorance, then I would say if the spectral intensity of green-yellow-orange in the middle is
(correction) equal to but not less than than the spectrum of blue, you get what appears when looking at the light is natural white light around 4000'K~4500'K
When cooler perceived colours are actually a hotter higher energy level of blue and thus the mean blackbody equivalent temperature rises above 5000'K meaning the ratio of Blue to Green-yelllow-red phosphors is higher.
I would offer a specification of 4500'K +/-250'K for optimum acuity and ensure that the range of intensity from between to peak levels is less than 10:1 in the main paths. <1% in all areas not needing light to avoid night blindness. It is not the intensity of Blue energy per se that is important but the distribution of BLue to Yellow ratio be balanced and the intensity be balanced.
The worst solution is glare where the lights themselves are visible. THis causes night blindness more than anything, so I would specify glare free ,<1% of peak reflected Lux on road surface when standing under any lamp looking towards any other lamp.
ALSO It is well known that art galleries prefer LED lights with lower deep blue spectral emission than HID lighting to reduce the aging effects of oil pigments from higher energy HID lighting that although has higher CRI also has more energy in UV and deep blue spectrum which is harmful to artwork.