[ODE] trimesh inertia

[David Walters]

> I'd like to contribute the function dMassSetTrimesh, based on this
> paper and code:
> http://jgt.akpeters.com/papers/Mirtich96/
> It worked very good in my own old physics engine.

Hi, I've commited your function to SVN, revision #1106, thanks for
working on it.

As a test, I've changed 'test_moving_trimesh' to use the new function
and behaviour in my limited tests appear to be correct.

Thanks again, it fills a noticable gap in functionality and it's
definitely good to not have to appoximate the mass of a mesh with a
box!

Regards,
David Walters