Fix test_weight_decay

[chengduoZH]

http://ci.paddlepaddle.org/viewLog.html?tab=buildLog&buildTypeId=Paddle_PrCi&buildId=89562

[22:51:27]	test_weight_decay failed
[22:51:27]	 F
[22:51:27]	======================================================================
[22:51:27]	FAIL: test_weight_decay (test_weight_decay.TestWeightDecay)
[22:51:27]	----------------------------------------------------------------------
[22:51:27]	Traceback (most recent call last):
[22:51:27]	  File "/paddle/build/python/paddle/fluid/tests/unittests/test_weight_decay.py", line 186, in test_weight_decay
[22:51:27]	    + " in class " + self.__class__.__name__)
[22:51:27]	AssertionError: Expect 0.00045114197
[22:51:27]	But Got0.00045114197 in class TestWeightDecay
[22:51:27]	
[22:51:27]	----------------------------------------------------------------------
[22:51:27]	Ran 1 test in 51.645s
[22:51:27]	
[22:51:27]	FAILED (failures=1)

Disable temporary, will fix it later.

[luotao1]

LGTM

[chengduoZH]

After analysis of this unit test, the reduce model doesn't support this usage:

            param_list = [(var, var * self.learning_rate)
                          for var in main_prog.block(0).all_parameters()]

            optimizer = fluid.optimizer.Adagrad(
                learning_rate=self.learning_rate)
            optimizer.minimize(avg_cost)

            for params in param_list:
                updated_p = fluid.layers.elementwise_sub(
                    x=params[0], y=params[1])
                fluid.layers.assign(input=updated_p, output=params[0])

Because, in Reduce mode, all the optimizer ops are scattered to different cards equally according to the op's role(which is opt_role) to update different parameter in a specified card, and after that, the updated parameters are broadcast to other cards. But the op_role of the above ops(below optimizer.minimize(avg_cost)) are forward, so the results are not right.