Change dropout api

[BBuf]

对齐Pytorch的dropout函数接口，交换generator和training参数位置。

[github-actions]

Speed stats:

GPU Name: GeForce GTX 1080 

OneFlow resnet50 time: 128.7ms (= 6435.4ms / 50, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 139.6ms (= 6982.0ms / 50, input_shape=[16, 3, 224, 224])
Relative speed: 1.08 (= 139.6ms / 128.7ms)

OneFlow resnet50 time: 74.6ms (= 3730.9ms / 50, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 88.0ms (= 4401.6ms / 50, input_shape=[8, 3, 224, 224])
Relative speed: 1.18 (= 88.0ms / 74.6ms)

OneFlow resnet50 time: 50.8ms (= 2537.9ms / 50, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 61.4ms (= 3070.0ms / 50, input_shape=[4, 3, 224, 224])
Relative speed: 1.21 (= 61.4ms / 50.8ms)

OneFlow resnet50 time: 45.8ms (= 2287.6ms / 50, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 48.8ms (= 2440.0ms / 50, input_shape=[2, 3, 224, 224])
Relative speed: 1.07 (= 48.8ms / 45.8ms)

OneFlow resnet50 time: 44.9ms (= 2244.2ms / 50, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 41.6ms (= 2077.6ms / 50, input_shape=[1, 3, 224, 224])
Relative speed: 0.93 (= 41.6ms / 44.9ms)

OneFlow resnet50 time: 153.8ms (= 7691.5ms / 50, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 158.9ms (= 7945.0ms / 50, input_shape=[16, 3, 224, 224], ddp, world size=2)
Relative speed: 1.03 (= 158.9ms / 153.8ms)

OneFlow resnet50 time: 102.7ms (= 5133.7ms / 50, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 100.8ms (= 5041.6ms / 50, input_shape=[8, 3, 224, 224], ddp, world size=2)
Relative speed: 0.98 (= 100.8ms / 102.7ms)

OneFlow resnet50 time: 83.3ms (= 4166.6ms / 50, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 73.7ms (= 3684.1ms / 50, input_shape=[4, 3, 224, 224], ddp, world size=2)
Relative speed: 0.88 (= 73.7ms / 83.3ms)

OneFlow resnet50 time: 75.5ms (= 3773.8ms / 50, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 62.3ms (= 3113.3ms / 50, input_shape=[2, 3, 224, 224], ddp, world size=2)
Relative speed: 0.82 (= 62.3ms / 75.5ms)

OneFlow resnet50 time: 67.3ms (= 3362.7ms / 50, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 58.9ms (= 2944.4ms / 50, input_shape=[1, 3, 224, 224], ddp, world size=2)
Relative speed: 0.88 (= 58.9ms / 67.3ms)