IobEncoder is not suitable for encoding SemArg annotations

[maxxkia]

Current implementation of IobEncoder is not suitable for encoding annotations of type SemArg.

IobEncoder requires an annotation type and a feature defined on that type. However, features of a SemArg are usually stored in a SemArgLink feature array (e.g. see TigerXmlReader, readSem() method, lines 330--364). Since the features are not stored in SemArg it makes it impossible to pass them to IobEncoder.