I agree. The codes tell you to give a color perception test but don't tell you what to do with it.
I know a few NDT hands that are either red or green color deficient, but I feel aslong as they can differentiate
between contrasts, they are ok. But it should be documented.

Being color deficient is not a death sentence to certification. Between the various forms of contrast ( luminous, subject, chromatic, Achromatic etc) the important ones for the color deficient individual will be luminous and achromatic. luminous contrast will be differentiating the background luminance vs the foreground luminance, and achromatic is the inverse of chromatic, in that it's differentiating between various shades of grey. For instance, if the tech failed to process a PT properly, leaving a background of what would normally be perceived as excessively pink, to the color blind individual, the entire area would run together and significantly reduce his or her ability to determine an indication vs. background as for that specific photopic (color day vision) response would be in the scotopic (night vision achromatic) range for those specific shades.

Luminous contrast therefore becomes of higher importance to this individual vs. someone with normal vision. The color deficient individual utilizes what you and I would consider mesopic vision which is that vision in-between photopic and scotopic.

To give you an idea, of the lighting levels in question (varies depending on the source, but it's roughly as follows) :
photopic vision is typically in the range of 1 to 10*6 cd/m*2 squared. (10 to the power of 6 cannot input that in this forum so I will use * to denote it)
mesopic vision occurs between 10*-6 up to 1 cd/m*2
Scotopic vision occurs between 10*-2 to 10*-6 cd/m*2
(btw, if your studying for the ASNT VT LIII, you better have those terms mermorized)
The color deficient inidivual should therefore be clearly in the photopic range to perform any chromatic contrast/luminous contrast dependent examination for the following reasons:

The visible spectrum extends from 380nm to approximately 770nm range. Drawing a line between those points, and raising the center line to the normal peak response somewhere around 550nm, it puts the peak response in the green range. But where is the red/orange? It's up at the top of the scale around 620 to 750nm. If that person finds themselves edging into the mesopic range, (when the cones and rods of the eye are working at the same time) They have lost a considerable amount of information and rather than having a limited amount of contrast as a normal vision would have, they have also lost the right and left sides of the spectral response that would allow them to utilize luminous contrast to fill in the blanks their deficiency left.

Going to the codes, this is part of the reason ASME defines specific lighting levels in Section V.
The illumination engineering society states that for general inspections, 50 to 100 foot candles, and for critical inspections 100 to 300 foot candles should be used. This is part of the reason you don't see specific information in regards to how to handle color deficient persons rather they only specify testing. If that person is firmly in the photopic range, their brain is going to extrapolate the missing information from either side of the scale, and insert achromatic (grey scale) information.

Given that the red end of the spectrum is the first to bite the bullet when the light level drops, you can understand the importance of why lighting becomes more critical for someone who is color deficient.

This also brings up another point I consider important. That person that is color deficient in the red range should not be using an LED for lighting.
There is not specific prohibition to this (yet) but they are setting themselves up for failure by doing so. 
There are some LEDs that emit standard white light, but they are very expensive. The most common (and cheapest) form of manufacturing of LED 'white' lights is coating an InGaN substrate (without the coating it's a pure blue light) with a phosphor coating. If you hold up the cheaper models to a white surface, you will note they appear blue tinged. This is the reason for that.

For this particular brand of LED the color rendering index (CRI) is achieved via stokes shift. Meaning that part of the blue light is transformed from shorter wavelengths to longer wavelengths.

The end result is the peak response level for this kind of LED is around 400 to 450nm dropping off significantly towards the bottom of the 500nm range and peaking back up around 550 to 600nm and trailing off dramatically towards the 700nm range which so happens to be the red end of the spectrum.

To sum that up, the spectral response of an InGaN LED is significantly reduced in the red (700nm) range vs the standard response of a filament light or sunlight.

If your color deficient in that range, and your attempting to utilize this kind of LED you've not only reduced your available spectral response to adjust for that deficiency, you've also inherently reduced your luminance contrast and set yourself up for failure.

The fix is a lot less complicated than the theory. Use a filament light instead of an LED if your color deficient.

There is nothing inherently wrong with a mild color deficiency as it can be adjusted for, but there are specific considerations that need to be addressed before cutting them loose to perform an exam.