Text this: VolumeDiffusion: Feed-forward text-to-3D generation with efficient volumetric encoder