I don't know of any technical (or declared) limits on the number of Hyperflex Clusters in one UCS Fabric Interconnect domain.
You may run into management limits by putting multiple independent Hyperflex clusters in one UCS Fabric Interconnect cluster.
(Think blast radius or shared fate failure domain.)
It brings up the question if you using one vCenter or multiple independent vCenter instances.
(Again think shared fate failure scenarios.)