Experimentally quantifying the advantages of weak-values-based metrology