Clarify how to convert `dok` dict from `bws.count` into a list of pairwise comparions

[ulf1]

the dok dict basically contains a sparse matrix, i.e. it matters if the indicies are (id1, id2) (i.e. Count(id1>id2)) or (id2, id1) (i.e. Count(id2>id1)).

other scoring models require inputs such as

(id1, id2, value=Bool(Count(id1>id2) > Count(id2>id1)), ...