[ODE] TriMesh and double

[Jeffrey Smith]

> I am trying to compile ode with TriMesh and double precision under 
> Linux, but the Makefile tells me this is impossible:

The only place in ODE that you MUST use floats is when feeding the vertex 
locations to OPCODE. Specifically, when using dGeomTriMeshDataBuild*. As
long as you're sure your vertex array is composed of floats, you can safely
disable that warning in the Makefile (it's around line 76, I beleive).  

-jeff