Wraps a rank N dataset, slicing on an index of the first dimension to make a rank N-1 dataset. This is currently used to implement DataSetOps.slice0(). Slicing a rank 1 dataset results in a rank 0 dataset. Supports rank 2 depend_1 datasets. Supports CONTEXT_0, DELTA_PLUS, DELTA_MINUS Supports BINS_1, JOIN_0