Earlier quoted context omitted.
In the email I received, it's a non-ASCII apostrophe encoded in UTF-8. The email headers say "Content-Type: text/html; charset=iso-8859-1", but then the HTML body says " ". Clients that give preference to the header will get it wrong. The full raw email: https://lukeshu.com/dump/startcom-email.txt
eh, the header should match the body encoding regardless of meta tag; should it not?
In HTML4/XHTML1, the isn't meant to be interpreted by the user agent; the HTTP server is supposed to parse it and set the the HTTP header accordingly.
Of course, the people writing the (X)HTML probably have a better idea of what encoding they're using than the people configuring the HTTP server, so it's common for user agents to allow to override the actual header, as it allows more things to work "correctly" for users. But, strictly speaking, that is non-conforming.
In (X)HTML5, that practice was codified, and the tag is given preference over the actual headers (for a whitelist of allowed headers).
So, which interpretation is "correct" depends on if it is HTML4/XHTML1 or (X)HTML5.
Now, the MIME type of the email body said "text/html" which can be anything, and the body used the HTML5 doctype, but specified the XHTML1 xmlns. I'm honestly not sure which interpretation is correct in that case.