btUniformScalingShape::calculateLocalInertia potentially incorrect

[akbyrd]

I noticed that the inertia calculation comes out wildly different when scaling a lone btConvexHullShape and when wrapping it with a btUniformScalingShape and scaling that instead.

In btUniformScalingShape::calculateLocalInertia it multiplies the inertia by the uniform scale. However, inertia is proportional to r^2 and thus it should be multiplying by scale^2. This should be analytically correct.

void btUniformScalingShape::calculateLocalInertia(btScalar mass, btVector3& inertia) const
{
	// this linear upscaling is not realistic, but we don't deal with large mass
	// ratios...
	btVector3 tmpInertia;
	m_childConvexShape->calculateLocalInertia(mass, tmpInertia);
	inertia = tmpInertia * m_uniformScalingFactor;
}