[ODE] nan

[Bram Stolk]

Hi,

I'm seeing NaN values in the unused parts of the rotation matrices.
dMatrix4 is a dReal R[3][4] where R[0][3], R[1][3], R[2][3] are not used.
The manual says:

 >The extra "padding" elements at the end of each row/column must be set to 0.
 >This is called the "standard format".

dRSetIdentity() sets these zero's.

In test_cyl, I use dRFromAxisAndAngle()
This function does not set the padding to zero, causing the NaN values.

I think we should set them to zero though.
It makes dRFromAxisAndAngle() a little slower, but at least it conforms
to spec, and NaN's are very nasty.
There are probably more funcs that skip the init.

   Bram