All members of a colonial animal are genetically identical, which I believe is why a man of war is an animal but an ant colony is a society. I don't think ants are genetically identical, although I'm not certain.
Man o Wars are also not disjoint. They have specialized cells, but I believe share nutrition among the non-feeding cells. A truly disjoint creature would need a metabolic system on both sides of the disjoint, a separate nervous system, and perhaps most difficult, a way to pass messages across the disjoint quickly enough to be useful (aka electric impulse).
At a certain point, I think it starts becoming dubious whether it's a single creature. Even if they're genetically identical, they start to look more like twins than a colonial creature.
Workers in a given ant colony - really, daughters of the same mother in any hymenopteran species, all of which use haplodiploid sex determination - are more closely related than siblings in species with fully diploid sex-determination systems, but they're not clones.
Colonial animals are the edge case here. Most would be hesitant to say an ant colony is "an animal", but are happy to refer to a Portuguese Man 'o' War as such, despite also being a colony.