Okay, today, the issue seems to have gone away.
I removed the utf8_encode call as well ensured the raw body was used without anything. Now all data formats are working as intended.
Thank you for the help and for pointing out the `byte-to-byte` comparison, which solved the issue.