Parallel pipeline planning not dependent on json_prolog

[bbferka]

• input and output types of annotators are extracted in the yaml parser and saved
• these are then used to plan the parallel execution of a pipeline;
• removed annotators and base class for classification;