fix: MIME-Encoded-Word-Betreffs mit Windows-1252 werden nicht dekodiert

decodeMIMEHeader() nutzte mime.WordDecoder ohne CharsetReader. Go kennt
dort nativ nur UTF-8/US-ASCII/ISO-8859-1, alles andere (z.B.
Windows-1252) ließ DecodeHeader() scheitern, Betreff blieb roh
("=?Windows-1252?Q?...?="). CharsetReader ergänzt, nutzt dieselbe
htmlindex-Logik wie decodeCharset() für den Body (vgl. PROJ-57).

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01WapWkrQusDuBMhaN8WyuXB

pkg/mailparser/parser.go

@@ -324,9 +324,20 @@ func decodeBody(data []byte, cte string) []byte {
 	return data
 }
 
-// decodeMIMEHeader decodes RFC 2047 encoded-word headers.
+// decodeMIMEHeader decodes RFC 2047 encoded-word headers. Go's mime package
+// only knows UTF-8/US-ASCII/ISO-8859-1 natively; CharsetReader closes that
+// gap for charsets like Windows-1252 using the same htmlindex lookup as
+// decodeCharset for the body.
 func decodeMIMEHeader(s string) string {
-	dec := new(mime.WordDecoder)
+	dec := &mime.WordDecoder{
+		CharsetReader: func(charset string, input io.Reader) (io.Reader, error) {
+			data, err := io.ReadAll(input)
+			if err != nil {
+				return nil, err
+			}
+			return bytes.NewReader(decodeCharset(data, charset)), nil
+		},
+	}
 	decoded, err := dec.DecodeHeader(s)
 	if err != nil {
 		return s