When running experiments, we collect some standard metrics for every algorithm, such as algo_name, node_expansions, runtime etc. But different algorithms also track other metrics, which may be unique to that method. We need a way of consolidating all of these metrics together and specifying on the command line which ones to print.