Determining default terminal colours is difficult and impossible in some cases. We could try to guess, but for the sake of simplicity it might be easier to do the following:

• Change white to a colour that works with most common background colours by default (green?)
• Add option to change colour scheme of the table, an optional dict perhaps e.g.:

colour_scheme = {
   default:'red',
   matched:'blue',
   underlined:'green'    
}
assert_df_equality(df1, df2, underline_cells=True, colour_scheme=colour_scheme)

We would need to add a validation step for the colour scheme dict.