sre 2.1b2 update:

- take locale into account for word boundary anchors (#410271)
- restored 2.0's *? behaviour (#233283, #408936 and others)
- speed up re.sub/re.subn

Lib/test/test_sre.py

@@ -329,6 +329,8 @@ for t in tests:
                 u = unicode(s, "latin-1")
             except NameError:
                 pass
+            except TypeError:
+                continue # skip unicode test strings
             else:
                 result=obj.search(u)
                 if result==None: