I find these definitions quite confusing and contrary to my current understanding of cosine loss. The term cosine gain makes no sense to me.
To define the term cosine loss I would say something like:
Cosine loss is the loss of power dependent on the angle between the downwind vector and the angle of the tether. Though not strictly a loss, cosine loss describes producing less power than would be possible directly downwind.
(the tether is considered straight in this context).
If the angle described above is \alpha, the cosine loss near the power zone may be approximated by multiplying the C_L by a factor \cos{\alpha}
A typical method of finding \alpha would be using the dot product of said vectors.
Another factor not taken into account here is the fact that in the power zone, yaw direction of the kite has no importance. When you travel away from the power zone, yaw will make a (sometimes slight) difference. Whether this effect is part of the term “cosine loss” I don’t know.
At the edge of the wind window, the cosine loss would be either complete or very big.
The term cosine loss seems to be quite loosely defined, and the use of the term is just to state that the kite is not working at full power directly downwind, rather at an angle, and thus full power is not available for harvesting. The bigger the angle, the more loss relative to working directly downwind.
I would not use the term in a mathematical exact sense.