find more accurately which files need to be doctested

[doombeaker]

现有根据 py 文件中是否又 "doctest" 来作为是否测试的标准，有些情况下容易误判，比如：

>>> output #doctest: +ELLIPSIS

为了能够愉快的使用 #doctest 一类的命令，改为根据 import doctest 判断，这和 https://github.com/Oneflow-Inc/oneflow/blob/master/ci/check/run_license_format.py#L38 这里的判断是一致的，更合理。

[github-actions]

Speed stats:

GPU Name: GeForce GTX 1080 

PyTorch resnet50 time: 139.2ms (= 6957.5ms / 50, input_shape=[16, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 128.4ms (= 6419.8ms / 50, input_shape=[16, 3, 224, 224], backward is enabled)
Relative speed: 1.08 (= 139.2ms / 128.4ms)

PyTorch resnet50 time: 83.1ms (= 4154.5ms / 50, input_shape=[8, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 74.7ms (= 3735.5ms / 50, input_shape=[8, 3, 224, 224], backward is enabled)
Relative speed: 1.11 (= 83.1ms / 74.7ms)

PyTorch resnet50 time: 58.4ms (= 2921.9ms / 50, input_shape=[4, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 47.6ms (= 2381.0ms / 50, input_shape=[4, 3, 224, 224], backward is enabled)
Relative speed: 1.23 (= 58.4ms / 47.6ms)

PyTorch resnet50 time: 47.7ms (= 2386.1ms / 50, input_shape=[2, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 38.2ms (= 1911.1ms / 50, input_shape=[2, 3, 224, 224], backward is enabled)
Relative speed: 1.25 (= 47.7ms / 38.2ms)

PyTorch resnet50 time: 45.1ms (= 2254.1ms / 50, input_shape=[1, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 41.2ms (= 2059.7ms / 50, input_shape=[1, 3, 224, 224], backward is enabled)
Relative speed: 1.09 (= 45.1ms / 41.2ms)

[github-actions]

Speed stats:

GPU Name: GeForce GTX 1080 

PyTorch resnet50 time: 141.5ms (= 7077.4ms / 50, input_shape=[16, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 128.4ms (= 6420.6ms / 50, input_shape=[16, 3, 224, 224], backward is enabled)
Relative speed: 1.10 (= 141.5ms / 128.4ms)

PyTorch resnet50 time: 84.5ms (= 4223.2ms / 50, input_shape=[8, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 74.3ms (= 3716.4ms / 50, input_shape=[8, 3, 224, 224], backward is enabled)
Relative speed: 1.14 (= 84.5ms / 74.3ms)

PyTorch resnet50 time: 58.7ms (= 2935.3ms / 50, input_shape=[4, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 50.4ms (= 2521.4ms / 50, input_shape=[4, 3, 224, 224], backward is enabled)
Relative speed: 1.16 (= 58.7ms / 50.4ms)

PyTorch resnet50 time: 48.4ms (= 2417.7ms / 50, input_shape=[2, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 44.7ms (= 2233.3ms / 50, input_shape=[2, 3, 224, 224], backward is enabled)
Relative speed: 1.08 (= 48.4ms / 44.7ms)

PyTorch resnet50 time: 42.3ms (= 2115.4ms / 50, input_shape=[1, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 40.2ms (= 2009.2ms / 50, input_shape=[1, 3, 224, 224], backward is enabled)
Relative speed: 1.05 (= 42.3ms / 40.2ms)