Triplets.collect returns incorrect values

[jegonzal]

The spark RDD.collect operation stores the output directly into an array. Since we reuse the iterator values only a single edge triplet is stored (in duplicate) for each partition.