bpo-39487: Merge duplicated _Py_IDENTIFIER identifiers in C code (GH-18254)

Moving repetitive `_Py_IDENTIFIER` instances to a global location helps identify them more easily in regards to sub-interpreter support.
diff --git a/Objects/bytesobject.c b/Objects/bytesobject.c
index f9823f1..5fd92f7 100644
--- a/Objects/bytesobject.c
+++ b/Objects/bytesobject.c
@@ -25,6 +25,8 @@
 static PyBytesObject *characters[UCHAR_MAX + 1];
 static PyBytesObject *nullstring;
 
+_Py_IDENTIFIER(__bytes__);
+
 /* PyBytesObject_SIZE gives the basic size of a string; any memory allocation
    for a string of length n should request PyBytesObject_SIZE + n bytes.
 
@@ -543,7 +545,6 @@
 format_obj(PyObject *v, const char **pbuf, Py_ssize_t *plen)
 {
     PyObject *func, *result;
-    _Py_IDENTIFIER(__bytes__);
     /* is it a bytes object? */
     if (PyBytes_Check(v)) {
         *pbuf = PyBytes_AS_STRING(v);
@@ -2485,7 +2486,6 @@
     PyObject *func;
     Py_ssize_t size;
     static char *kwlist[] = {"source", "encoding", "errors", 0};
-    _Py_IDENTIFIER(__bytes__);
 
     if (type != &PyBytes_Type)
         return bytes_subtype_new(type, args, kwds);