#277 PRMS performance with maxscore

[fsqcds]

The bug is similar to https://sourceforge.net/p/lemur/bugs/263/ and has the same workaround:--processingModel=rankeddocument

Right now I on IMDB dataset dataset with PRMS operator and mu=90 am getting map=0.433 with maxscore and map=0.560 with rankeddocument. Performance of other models I've tried (QL, SDM, WSDM) doesn't change.