Add ceil_mode option for pool_op

[wanghaoshuang]

Add ceil_mode option the same as V2 API:
https://github.com/PaddlePaddle/Paddle/blob/develop/python/paddle/trainer_config_helpers/layers.py#L2746

[wanghaoshuang]

• tensorflow pool:
  不支持explicit padding ， 只提供了padding option用来选择 padding algorithm： "SAME" or "VALID".

If padding = "SAME": output_spatial_shape[i] = ceil(input_spatial_shape[i] / strides[i])
If padding = "VALID": output_spatial_shape[i] = ceil((input_spatial_shape[i] - (window_shape[i] - 1) * dilation_rate[i]) / strides[i]).

padding = "SAME" 相当于 paddle中padding=None and ceil_mode=True
padding = "VALID" 相当于paddle中padding=None and ceil_mode=False

tensorflow这种方法无法兼容V2 OCR CTC model (padding=1 and ceil_mode=True)

• onnx:
  onnx 关于pool op没有直接的定义，但是可以参考onnx conv的定义。

类似于tensorflow, 提供auto_pad option:

auto_pad must be either SAME_UPPER, SAME_LOWER or VALID. Where SAME_UPPER or SAME_LOWER mean pad the input so that the output size match the input.In case of odd number add the extra padding at the end for SAME_UPPER and at the beginning for SAME_LOWER. VALID mean no padding. DEPRECATION NOTE: auto_pad is only intended to support legacy uses, and for framework authors, one is explicitly encouraged to use explicit padding specified in the pads attribute.

onnx 没有定义output size计算方式。

• 总结：
  考虑到tensorflow pool op的限制和onnx的建议，explicit padding + ceil_mode 貌似是比较好的选择。