*: Use default argument for str.encode()

I had missed that this became hard-coded to utf-8 in py3k,
instead of being overridable as in python 2.

Automated change with modified pyupgrade.

tornado/test/escape_test.py

@@ -247,7 +247,7 @@ def test_xhtml_unescape_numeric(self):
     def test_url_escape_unicode(self):
         tests = [
             # byte strings are passed through as-is
-            ("\u00e9".encode("utf8"), "%C3%A9"),
+            ("\u00e9".encode(), "%C3%A9"),
             ("\u00e9".encode("latin1"), "%E9"),
             # unicode strings become utf8
             ("\u00e9", "%C3%A9"),

tornado/test/httpserver_test.py

@@ -282,13 +282,11 @@ def test_multipart_form(self):
                 [
                     b"Content-Disposition: form-data; name=argument",
                     b"",
-                    "\u00e1".encode("utf-8"),
+                    "\u00e1".encode(),
                     b"--1234567890",
-                    'Content-Disposition: form-data; name="files"; filename="\u00f3"'.encode(
-                        "utf8"
-                    ),
+                    'Content-Disposition: form-data; name="files"; filename="\u00f3"'.encode(),
                     b"",
-                    "\u00fa".encode("utf-8"),
+                    "\u00fa".encode(),
                     b"--1234567890--",
                     b"",
                 ]

tornado/test/log_test.py

@@ -89,7 +89,7 @@ def test_bytes_logging(self):
 
     def test_utf8_logging(self):
         with ignore_bytes_warning():
-            self.logger.error("\u00e9".encode("utf8"))
+            self.logger.error("\u00e9".encode())
         if issubclass(bytes, basestring_type):
             # on python 2, utf8 byte strings (and by extension ascii byte
             # strings) are passed through as-is.