Quite complex algorith is being applied to list of Spark Dataset\'s rows (list was obtained using groupByKey and flatMapGroups). Most rows are transformed 1 : 1 from input t