This function is called by RankAggregator::consensusRanking.
The heuristic procedure orders unranked Items according the proportion
of times an item was preferred in all pairwise comparisons with other unranked
Items.
extendRanking(umat, node)a matrix, which is either the output of evaluationMatrix,
or a subset of the output of evaluationMatrix.
a list of elements, containing information about a node in
the branch and bound search space. The relevant elements here are
$partial.ranking, $included, and $prl. Where,
$partial.ranking is a vector of rank positions for each Item
in umat that is ranked so far; partial rankings may contain some -
or all - NA values. $included is a logical vector denoting if
an Item in umat is ranked in $partial.ranking. And
$prl is a numeric value denoting how many of the Items in
umat are ranked in $partial.ranking.
A vector of rank positions.