On insert of a model store the type of the model in the document

[jettro]

At the moment you would have to parse the model string to obtain the type of the model used by the ranklib library (i.e. LambdaRank, ListNet). As mentioned in slack by @nomoa the model is parsed on insertion. So theoretically we could deduce the type of model in that phase and insert is as sort of a meta data field in the document. That way obtaining the models and explaining the type of the used model in a generic tool becomes a lot easier.