The description of ResolvedJavaType.hasFinalizableSubclass should be rephrased since it returns an AssumptionResult<Boolean> rather than a boolean.
Maybe a provision for returning null when the JVM doesn't want to make any promises either way and the compiler should take the conservative option.