I'm happy to hear you're considering helix as well; I think it's the same in this context when unwrapped as you pointed out.

Back to the math. Here is what I had in mind. in terms of finding a feed rate for a ramp:

I see two approaches. I called them the "linear" solution and the "elliptical" solution. I favor the elliptical approach. It sounds like @chrisb is leaning toward linear approach.

When I solved the simultaneous equations, I could not eliminate the ugly tangent functions. I'm not a big fan because they require special handling at the extremes. I'm sure a math enthuisiast could come up with a much more elegant solution....

It boils down to these two

Linear solution:

Vx = Vzmax / ( tan(theta) + ( Vzmax/Vxmax ) )

Vz = Vx tan(theta)

Elliptical solution:

Vx = 1 / sqrt( (tan(theta)^2 / Vzmax) + 1/ Vxmax^2 )

Vz = Vx tan(theta)

Where:

theta = ramp angle

tan(theta) = Zcut / Xcut

Vx and Vz are x and z components of the feed vector respectively.

Vxmax and Vzmax are the set horizontal and vertical feed rates respectively.

If anybody would like to laugh at my chicken scratch arriving at these, here they are:

I also made an Excel spreadsheet to test out that these schemes aren't completely nuts.

There are probably more elegant solutions. I'm even convinced that the original solution (pick one), @chrisb's conditional formula, linear and elliptical solutions can probably all be generalized into one equation with the addition of another "how daring are you?" constant. The words p-norm and Lp-space come to mind, but those are way outside my understanding... Did I get any math enthusiasts excited?