Sparse sketching: SASO and LASO or MajorAxis::Short and MajorAxis::Long

[rileyjmurray]

We recently introduced the .major_axis member to the DenseSkOp struct definition, which can take values in the MajorAxis enumeration. The concept of a major axis also arises in sparse sketching operators. We've been using the phrase "short-axis sparse operator" and "long-axis sparse operator" for what we might otherwise call operators with short and long major axes, respectively. I'm wondering if it makes sense to unify the terminology here and replace SparseSkOp.family with SparseSkOp.major_axis, and replace SASO/LASO with MajorAxis::Short and MajorAxis::Long. One benefit of this is that we could replace SparseSkOp.vec_nnz with SparseSkOp.major_axis_vec_nnz, which has more of a "self-documenting" flavor.