Add an option to define the `itersize` of the server-side cursor

Large imports depend on the use of server-side cursors as otherwise out of memory errors may occur. The use of this type of cursor needs a parameter specifying how many datasets are fetched per call. Because this number depends on the memory on the client side, it is beneficial to make it configurable for the user to have the value as high as possible for importing speed.

EDIT: it turned out no implementation is needed, but the documentation of the options was updated.

\rfc @tara @shinde @laurens @gislars