fix python example: grid_sampler, kldiv_loss, spectral_norm (#1754)

doc/fluid/api_cn/layers_cn/grid_sampler_cn.rst

@@ -62,6 +62,8 @@ step 2：
 
 .. code-block:: python
 
+    import paddle.fluid as fluid
+
     # 一般与 affine_grid 组合使用
     x = fluid.data(name='x', shape=[None, 10, 32, 32], dtype='float32')
     theta = fluid.layers.data(name='theta', shape=[2, 3], dtype='float32')

doc/fluid/api_cn/layers_cn/kldiv_loss_cn.rst

@@ -34,10 +34,12 @@ kL发散损失计算如下：
 
 返回类型：变量(Variable)，数据类型与输入 ``x`` 一致。
 
-**代码示例**：
+**代码示例：**
 
 .. code-block:: python
 
+    import paddle.fluid as fluid
+
     # 'batchmean' reduction, loss shape 为[N]
     x = fluid.data(name='x', shape=[None,4,2,2], dtype='float32') # shape=[-1, 4, 2, 2]
     target = fluid.layers.data(name='target', shape=[4,2,2], dtype='float32')

doc/fluid/api_cn/layers_cn/spectral_norm_cn.rst

@@ -37,10 +37,12 @@ spectral_norm
 
 返回类型：变量(Variable)，数据类型与输入 ``weight`` 一致。
 
-**代码示例**：
+**代码示例：**
 
 .. code-block:: python
 
+    import paddle.fluid as fluid
+
     weight = fluid.data(name='weight', shape=[2, 8, 32, 32], dtype='float32')
     x = fluid.layers.spectral_norm(weight=weight, dim=1, power_iters=2)