Batched parallel not working with UNWIND

[katarinasupe]

Import the following dataset with batched parallel in-memory analytical mode and notice you have 300k nodes and 280k edges, while the dataset has 300k nodes and 300k edges (can be confirmed with classic mgconsole import with in-memory transactional mode). To be exact, there are 20k relationships of type DRIVES missing.

Here is the dataset.

[katarinasupe]

Probably related to UNWIND clause being used.
Tested on just nodes import and again ended up with 20K nodes less than it should be (in this case, Order nodes) both in in-memory analytical and in-memory transactional.

Here is the dataset.