3rd solution for #734

void
MSet::Internal::set_new_weights(vector<double>::iterator weights_begin,vector<double>::iterator weights_end) {
    max_attained = INT_MIN;
    while (weights_begin!=weights_end)
        items[i].wt = *weight_begin;
        max_attained = max(max_attained,items[i].wt);
        weight_being++;
        // Dont know what the max_possible should be.
    }
}

template <typename Iterator>
void
MSet::set_new_weights(Iterator first, Iterator last) {
   set_new_weights_helper(first,last);
}

void
Mset::set_new_weights_helper(vector<double>::iterator weights_begin,vector<double>::iterator weights_end)
{
    internal->set_new_weights(weights_being,weights_end);
}

void
Ranker::rank(Xapian::MSet & mset, const string & model_key, const Xapian::FeatureList & flist)
{
    LOGCALL(API, std::vector<Xapian::docid>, "Ranker::rank", mset | model_key | flist);
    std::vector<FeatureVector> fvv = flist.create_feature_vectors(mset, letor_query, Xapian::Database(db_path));
    load_model_from_metadata(model_key);
    std::vector<FeatureVector> rankedfvv = rank_fvv(fvv);/*Not sorted, The FeatureVectors are in the original order of the  documents*/

    std::vector<double> weights;
    for (size_t i = 0; i < rankedfvv.size(); ++i) {
        weights.push_back(rankedfvv[i].get_score());
    }

   mset.set_new_weights(weights);
}
/* The returned weights can be linearly combined for multiple rankers and then we can call MSet::set_new_weights_and_rerank.