add shape not match doc to data layer, test=develop

Author: sneaxiy

paddle/fluid/API.spec

@@ -238,7 +238,7 @@ paddle.fluid.layers.continuous_value_model (ArgSpec(args=['input', 'cvm', 'use_c
 paddle.fluid.layers.where (ArgSpec(args=['condition'], varargs=None, keywords=None, defaults=None), ('document', '3126e3039e752ce26077f1efaca355c6'))
 paddle.fluid.layers.sign (ArgSpec(args=['x'], varargs=None, keywords=None, defaults=None), ('document', 'ccf6bb7912afd2818d24bc45461e807a'))
 paddle.fluid.layers.deformable_conv (ArgSpec(args=['input', 'offset', 'mask', 'num_filters', 'filter_size', 'stride', 'padding', 'dilation', 'groups', 'deformable_groups', 'im2col_step', 'param_attr', 'bias_attr', 'name'], varargs=None, keywords=None, defaults=(1, 0, 1, None, None, None, None, None, None)), ('document', 'c896b66265a60bd3c5510f66e6e02919'))
-paddle.fluid.layers.data (ArgSpec(args=['name', 'shape', 'append_batch_size', 'dtype', 'lod_level', 'type', 'stop_gradient'], varargs=None, keywords=None, defaults=(True, 'float32', 0, VarType.LOD_TENSOR, True)), ('document', '6e19128b46936edf9f3fad77860a1da8'))
+paddle.fluid.layers.data (ArgSpec(args=['name', 'shape', 'append_batch_size', 'dtype', 'lod_level', 'type', 'stop_gradient'], varargs=None, keywords=None, defaults=(True, 'float32', 0, VarType.LOD_TENSOR, True)), ('document', 'f012ab25126ce6d03d71ce80dc7e5e41'))
 paddle.fluid.layers.open_files (ArgSpec(args=['filenames', 'shapes', 'lod_levels', 'dtypes', 'thread_num', 'buffer_size', 'pass_num', 'is_test'], varargs=None, keywords=None, defaults=(None, None, 1, None)), ('document', 'dce69a78638da8f7ad80b1fc00ed2029'))
 paddle.fluid.layers.read_file (ArgSpec(args=['reader'], varargs=None, keywords=None, defaults=None), ('document', '32181f6037e387fb6e68a5beaafe33b6'))
 paddle.fluid.layers.shuffle (ArgSpec(args=['reader', 'buffer_size'], varargs=None, keywords=None, defaults=None), ('document', 'f29d7d159e114f73fc988d9a86805841'))

python/paddle/fluid/layers/io.py

@@ -54,6 +54,11 @@ def data(name,
     All the input variables of this function are passed in as local variables
     to the LayerHelper constructor.
 
+    Notice that paddle would only use :code:`shape` to infer the shapes of 
+    following variables in the network during compile-time. During run-time, 
+    paddle would not check whether the shape of the feeded data matches the 
+    :code:`shape` settings in this function. 
+
     Args:
        name(str): The name/alias of the function
        shape(list): Tuple declaring the shape. If :code:`append_batch_size` is