Slight modification of org.apache.spark.rdd.ZippedWithIndexRDD from Spark that does not run job when created, it only triggers job on action or computing partitions.
Wrapper for RDD.zipPartitions that fails early if the RDDs' partition-counts don't match