gh-155742: Use PyBytesWriter in codecs

Replace soft deprecated PyBytes_FromStringAndSize() with
PyBytesWriter.

Author: vstinner

Python/codecs.c

@@ -1426,26 +1426,27 @@ _PyCodec_SurrogateEscapeUnicodeEncodeError(PyObject *exc)
         return NULL;
     }
 
-    PyObject *res = PyBytes_FromStringAndSize(NULL, slen);
-    if (res == NULL) {
+    PyBytesWriter *writer = PyBytesWriter_Create(slen);
+    if (writer == NULL) {
         Py_DECREF(obj);
         return NULL;
     }
 
-    char *outp = PyBytes_AsString(res);
+    char *outp = PyBytesWriter_GetData(writer);
     for (Py_ssize_t i = start; i < end; i++) {
         Py_UCS4 ch = PyUnicode_READ_CHAR(obj, i);
         if (ch < 0xdc80 || ch > 0xdcff) {
             /* Not a UTF-8b surrogate, fail with original exception. */
             Py_DECREF(obj);
-            Py_DECREF(res);
+            PyBytesWriter_Discard(writer);
             PyErr_SetObject(PyExceptionInstance_Class(exc), exc);
             return NULL;
         }
         *outp++ = ch - 0xdc00;
     }
     Py_DECREF(obj);
 
+    PyObject *res = PyBytesWriter_Finish(writer);  // can be NULL
     return Py_BuildValue("(Nn)", res, end);
 }