The graviton is not part of GR; it is neither predicted nor required by the theory since, as you rightly say, no forces are involved. Gravitons arise only if we try to formulate a quantum field theory of gravity, which is a step beyond "classic" GR. Such a field theory would require a particle as mediator of the force, which is precisely the ( hypothetical ) graviton. Since the source term of gravity is a rank-2 tensor, the graviton would need to be a massless spin-2 boson. No, you need to be careful here - GR deals with intrinsic geometry, not extrinsic one. The latter would require an embedding of the manifold in a higher dimensional space, but the former doesn't. Of course there exist models where our universe is indeed embedded in a higher dimensional manifold ( with very interesting results ! ), but "classic" GR does not require this to be the case at all. There are no mechanical forces involved in this, just geometry.