on keyword vs text, keyword is really about docvalues usage whereas text is about inverted index. 5.1 plans to have an analyzer on keyword as long as there is no tokenization or rather multiple tokens output. Objective is lower case etc.