That an automatic conversion happens for different types but not
for different encodings was always a bit odd.
This makes writing tests easier and comparing values within the
tag editor does not rely on choosing a particular encoding.
* Convert UTF-8 text fields automatically to UTF-16
when version is ID3v2.3 or below
-> so conversion from ID3v2.4 to ID3v2.3 or below
doesn't require manual conversion of possibly
UTF-8 encoded fields
* Don't use magic numbers for ID3v2 encoding byte
* Don't warn about UTF-16 Big Endian with BOM