read_df confused by (useless in utf8) byte order marks

[fiendish]

See: #493 (comment)
And: https://docs.python.org/3/library/codecs.html#encodings-and-unicode

I think read_df should understand file encoding when reading files