Decode a response body by the charset its Content-Type names (UTF-8 when it names
none). TextDecoder drops a leading byte order mark, which Buffer#toString keeps and
JSON.parse then rejects, so a BOM added by a proxy or a backend change cannot turn a
valid answer into a parse error, and a Latin-1 body keeps its umlauts ("München").
An unknown charset label is a LadesaeulenParseError.
Decode a response body by the charset its Content-Type names (UTF-8 when it names none). TextDecoder drops a leading byte order mark, which Buffer#toString keeps and JSON.parse then rejects, so a BOM added by a proxy or a backend change cannot turn a valid answer into a parse error, and a Latin-1 body keeps its umlauts ("München"). An unknown charset label is a LadesaeulenParseError.