When showing a kanban, there is differences between dataset of types:
* DataSetStatic: self.view.dataset===self.dataset, their ids attributes are the entire ids list,
* DataSet and DataSetSearch: self.view.dataset.ids are the already in the view ids, self.dataset.ids are the last gotten ids.
Hence with DataSetStatic dataset, when self.view.dataset.ids.splice(0) is done
self.dataset.ids is also emptied. And in the read_slice function, the slice is
done on that (now empty) array.
This fix removes the splicing of this ids array (which doesn't change a thing
since the array is overwritten latter), a _.difference is used to remove
eventual duplicates since in the DataSetStatic case, the same array is being
concatenated to itself
opw-630654 opw-617090 opw-619563