xla

c54fb2a8 - Complete support for einsum via PyTorch lowering

Complete support for einsum via PyTorch lowering Just go directly to at::native::einsum to allow PyTorch lowering to use lazy tensors, we have already implemented the tensor methods used by it.