The API documentation seems to imply (but without being explicit) that there is a tree structure to the data tables, starting from the root-level tables with each having a distinct set of subTables and sub-sub-Tables.
But the API endpoint to link parent and child table could conceivably be used to form more complicated dependency graphs. One could for example have the same subTable included in more than one root table, thereby reusing field definitions. OTOH, this adds complexity. Cycles should certainly be avoided (but may be hard to detect).