Lazy Interpret support infer default device cpu

[chengtbf]

• Lazy Interpret 支持推导默认的 Device（当 Ctx 中的 placement、 device 均为空时，默认生成 cpu device）
• 增加 arange 单测并测试。

[leaves-zwx]

测试是否需要覆盖一下 device 未设置，使用默认 cpu device 的 case?

[github-actions]

Speed stats:

GPU Name: GeForce GTX 1080 

PyTorch resnet50 time: 142.0ms (= 7100.7ms / 50, input_shape=[16, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 128.0ms (= 6400.5ms / 50, input_shape=[16, 3, 224, 224], backward is enabled)
Relative speed: 1.11 (= 142.0ms / 128.0ms)

PyTorch resnet50 time: 83.8ms (= 4190.9ms / 50, input_shape=[8, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 74.3ms (= 3713.4ms / 50, input_shape=[8, 3, 224, 224], backward is enabled)
Relative speed: 1.13 (= 83.8ms / 74.3ms)

PyTorch resnet50 time: 55.9ms (= 2792.9ms / 50, input_shape=[4, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 47.1ms (= 2355.2ms / 50, input_shape=[4, 3, 224, 224], backward is enabled)
Relative speed: 1.19 (= 55.9ms / 47.1ms)

PyTorch resnet50 time: 48.6ms (= 2430.3ms / 50, input_shape=[2, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 38.9ms (= 1947.2ms / 50, input_shape=[2, 3, 224, 224], backward is enabled)
Relative speed: 1.25 (= 48.6ms / 38.9ms)

PyTorch resnet50 time: 43.3ms (= 2166.4ms / 50, input_shape=[1, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 37.1ms (= 1854.1ms / 50, input_shape=[1, 3, 224, 224], backward is enabled)
Relative speed: 1.17 (= 43.3ms / 37.1ms)