The loss of magnetism at the Curie temperature applies primarily to "induced magnetism like iron sticking to a magnet".
Ferrofluid doesn't really include a molten iron; ferrofluid is a collection of many small but mezoscopic particles, "sawdust", and its magnetism doesn't differ so much from magnetism of normal pieces of iron except that it's easier for the particles to change the orientation. Ferrofluid normally have lots of "normal fluid" in it, like water or organic liquids, and some "coating", so the ferromagnetic material is a relatively minor component and it's surely not melted.
When a ferrofluid gets heated to the Curie temperature, it loses its magnetism, too.
The Curie temperature of all sensible materials is and has to be below the melting point. When you approach the melting point while heating the material, the magnetism has been lost for quite some time.
Yes, molten iron is paramagnetic, much like every ferromagnetic material above the Curie temperature.