A small but functional demo program that loads a given RandomIndex and
generates some semantic sets, either for the given index terms or for the
highest tf*idf ranking index terms in the index.
Please note that TestRI is very naively implemented and does _not_
demonstrate recommended use of Random Indexes. Typically you do not, as
TestRI does, want to compare all representations (vectors) with each other,
but rather compare specific tokens with a limited set of tokens in order
to get a ranked list of the set. Random Indexes tend to grow big.