Fix: Unique facets now correctly returns only the most specific labels
Restores subclass semantics in getSublabelMapping() which was broken in v0.0.2. The switch to part-of relationship checking caused the uniqueness filter to be bypassed entirely — all labels were returned rather than only the most specific ones.
Root cause
Label class expressions in the config relate to each other via subclass hierarchies (e.g. neuron is a subclass of cell). The reasoner found no entailments for part-of axioms, so every label's sublabel set remained empty, and the subLabels.isEmpty() condition in addUniqueNodeLabels() always evaluated to true, keeping all labels.
Fix
Reverted to SubClassOf axiom checking while retaining the equivalence guard (ce.equals(ceCompare)) introduced in v0.0.2.