[SPARK-1687] [PySpark] fix unit tests related to pickable namedtuple

serializer is imported multiple times during doctests, so it's better to make _hijack_namedtuple() safe to be called multiple times.

Author: Davies Liu <davies.liu@gmail.com>

Closes #1771 from davies/fix and squashes the following commits:

1a9e336 [Davies Liu] fix unit tests

python/pyspark/serializers.py

@@ -297,8 +297,11 @@ def _hack_namedtuple(cls):
 
 def _hijack_namedtuple():
     """ Hack namedtuple() to make it picklable """
-    global _old_namedtuple # or it will put in closure
+    # hijack only one time
+    if hasattr(collections.namedtuple, "__hijack"):
+        return
 
+    global _old_namedtuple # or it will put in closure
     def _copy_func(f):
         return types.FunctionType(f.func_code, f.func_globals, f.func_name,
                 f.func_defaults, f.func_closure)
@@ -313,6 +316,7 @@ def _hijack_namedtuple():
     collections.namedtuple.func_globals["_old_namedtuple"] = _old_namedtuple
     collections.namedtuple.func_globals["_hack_namedtuple"] = _hack_namedtuple
     collections.namedtuple.func_code = namedtuple.func_code
+    collections.namedtuple.__hijack = 1
 
     # hack the cls already generated by namedtuple
     # those created in other module can be pickled as normal,