Parser should check that valid unknown schema nodes are defined as language extensions

Currently if parser encounters a node that is not known, it is added to collection of unknown schema nodes. I think the parsing should me more strict and fail on unknown nodes that do not belong to a defined language extension.