The code here assumes that if dupRecords is passed, then the caller wants a set of un-inserted duplicates. If dupRecords is passed, the uniqueness of the index informs whether or not to ignore duplicates, overriding any previous call ignoreUniqueConstraint(). In this case x_1 is unique, so it will not ignore duplicates.

In this case, the caller doesn't even want this set of un-inserted duplicate records. We should just not pass dupRecords to dumpInsertsFromBulk for secondary indexes, so that all duplicates get inserted and ignored, as expected.