Well, I don’t know anything about calibrating scanners, and with that general disclaimer, I’d remind u that the Tristimulus Values are based on the Standard Observer when presented with the same illuminant and it’s only an estimate when all is said and done.
Same with the way the scanner ‘sees.’ It will not have a value differential. You need to convert X,Y,Z to an rgb space without value. That is the why I think u can’t find it on the easyrgb site. And as u say, the code is “neutral.” IF I understand you correctly! Big IF.
Try first converting to rgb Then, add the lab conversion. Cie only has an X, Y coordinate. X= x / by x + y =z; Y= y/x=y=z. ( z blue, y green, x red )

Take the scene that the scanner ‘sees’ at your D50 illuminant, and integrate it at each point along the graph and then sum them together.

Then u need to add the last dimension- value. L is the value dimension on a scale where 100 is White and 0 is Black. a is the ‘green’ scale ( a- = green; a+ =red), b is the ‘blue’ scale ( b-=blue; b+=yellow.) L is calculated by dividing the square root of lower case y by 10; a is 17.5 x (1.02x-y)/square root of y; and b=7(4-0.472)/square root of y.

You may want to consult one of the optical engineers at the RIT, Munsell Color Lab.
http://query.nytimes.com/gst/fullpage.html?res=9C06EEDA1F3BF935A15750C0A961958260&sec=health&spon=&pagewanted=2

There is also an instrument evaluation data option under Resources in the pull down menu.
Bonne Chance!