The problem described in !52816 (merged) turns out to be due to this function, where if no SG::BaseInfoBase is available for a clid it returns true, meaning that where we don't have base information we assume that everything inherits from IParticle. However, the main reason we do not have this information available is because no-one has called one of the DATAVECTOR_BASE macros, i.e. no-one has told us anything about bases. In these cases it seems safest to return false (i.e. where we don't have base information we assume no inheritance).
This should fix the crash seen in that other MR.