I have downloaded emails from Gmail via Google Takeout, one label to be exact. Looking closer at the downloaded contents (in MBOX format), it turns out that all non-ASCII characters have been replaced by the byte sequence
EF BF BD: the UTF-8 encoded Unicode REPLACEMENT CHARACTER (U+FFFD: �)!
I would hope there is a solution to that...?
Edit: The emails do not include any
Content-Type header, so Gmail cannot know which character set they contain. Gmail displays the emails just fine. It probably guesses that it's ISO-Latin-1, or whatever works best for the specific email. Takeout then just assumes UTF-8 and converts every weird character into the replacement character...
I have not found any way around this, using Takeout. Except making sure the original emails are in UTF-8, but that's not always in my control.