I’ve found the following counter-example, to this conjecture:
Example For a set $latex S$ of binary relations
$latex \forall X_0,\dots,X_n\in S:\mathrm{up}(X_0\sqcap^{\mathsf{FCD}}\dots\sqcap^{\mathsf{FCD}} X_n)\subseteq S$
does not imply that there exists funcoid $latex f$ such that $latex S=\mathrm{up}\, f$.
The proof is currently available at this PDF file and this wiki page.