Fix ElementTree.write XML prolog for encoding 'utf-8-sig'

When ElementTree object is to be written to the file, and when BOM
is needed, the 'utf-8-sig' can be used for the purpose.
However, the XML prolog then looks like...

<?xml version='1.0' encoding='utf-8-sig'?>

... and that encoding in the prolog makes no sense. Therefore,
the utf-8-sig is changed to utf-8 for the purpose.

Author: pepr

Lib/xml/etree/ElementTree.py

@@ -738,6 +738,8 @@ def write(self, file_or_filename,
                     # Retrieve the default encoding for the xml declaration
                     import locale
                     declared_encoding = locale.getpreferredencoding()
+                elif enc_lower == "utf-8-sig":
+                    declared_encoding = "utf-8"
                 write("<?xml version='1.0' encoding='%s'?>\n" % (
                     declared_encoding,))
             if method == "text":