Node that samples rows from an input table. The sampling is done either random or top rows first and the user needs also to specify whether the a given fractin (in [0,1]) or an absolute count of the input table's rows should survive.