Re-encode depth to tiled 4k

Take 6 720 depth images and tile into a single 4k depth image. Or take 8 and tile but with a 60 pixel loss around the edges. This would allow the already corrected and aligned depth maps, with redundancy removed, to be re-encoded. This allows full 3D viewing of the scene without needing to replicate the reconstruction process at the viewer. It should also result in improved compression due to greatly reduced noise and redundancy.

It requires a major rethink of how sources are represented and encoded. A single source can now contain a set of frames rather than just one and some tiling / de-tiling process is required. This is essentially needed anyway to properly store a scene.