df1 and df1 have the same entries and labels.
df1 and df2 have different entries.
However, they are pairwise in the relation float(2, digits=3) (see verify/float).
While df1 and df3 map the same labels to the same values, they are ordered differently (in particular, the rows are swapped). The verify/dataframe command has no facility to recognize this.
df1 and df4 have the same entries, but different column labels. We specify columnlabelverification = true (see verify/true) so that the labels compare pairwise successfully.