[ODE] Terrain<->Box collision without tri-collider

[gl]

(I actually missed this email, just stumbled on it in my inbox...)

> I also think this is the way to go, it makes IMHO no sense to compute the
normals again, then we have almost the same situation as with the
tri-collider, data and cpu time duplication.

Hmm, it would be nice to have both options.  In my app for example I can
generate terrain normals from the heightmap on the 3D board, so I don't need
to store them in memory for lighting.  On the other hand, some people may
already have vertex normals for lighting.  I guess it depends how heavy the
hit is from auto-computation, and how large your heightfields are (usual
memory vs CPU tradeoff).

> Ok, i actually haven't look at the license but i think finding out how the
SlickWorm author does his sphere<->heighmapt intersection and re-engineering
it would be ok.

Sure.

> Fabian, sounds great - love to look at the code (but I won't be able to
> spend time on it until next week).
--
gl