experiment with utf-8 encoding to see if efforts were wasted

rplugin/python3/handlers/chat_handler.py

@@ -229,7 +229,7 @@ def _add_chat_messages(
                 last_line_col,
                 last_line_row,
                 last_line_col,
-                token.split("\n"),
+                [tok.encode("utf-8") for tok in token.split("\n")],
             )
 
     def _add_end_separator(self, model: str, no_annoyance: bool = False):