Filing this in response to BF-31204. This test was timing out on two in-memory variants, so the immediate fix was to prevent the passthrough from running on those variants, presumably due to the in-memory variants being slower and timing out. However, although it doesn't appear that this test is timing out in other non in-memory variants, running this operation it does take a long time (potentially does time out). We should double check why this is happening - one theory is because the batch is twice the size (due to the implict insert override in the time series suite inserting a time field in every insert).