The autocovariance function for marked point processes: a comparison between two different approaches