Remove GpuDeviceNum

[liujuncheng]

移除 GpuDeviceNum 的使用，改为依赖实际的设备数量，避免和具体的设备类型耦合

[github-actions]

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/8166/

[github-actions]

Speed stats:

GPU Name: NVIDIA GeForce GTX 1080 

✔️ OneFlow resnet50 time: 129.0ms (= 12899.3ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 142.3ms (= 14233.3ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.10 (= 142.3ms / 129.0ms)

OneFlow resnet50 time: 76.6ms (= 7661.0ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 84.3ms (= 8427.1ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.10 (= 84.3ms / 76.6ms)

OneFlow resnet50 time: 51.6ms (= 10324.5ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 58.6ms (= 11729.9ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.14 (= 58.6ms / 51.6ms)

OneFlow resnet50 time: 43.3ms (= 8663.1ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 39.2ms (= 7834.7ms / 200, input_shape=[2, 3, 224, 224])
❌ Relative speed: 0.90 (= 39.2ms / 43.3ms)

OneFlow resnet50 time: 38.7ms (= 7748.6ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 43.8ms (= 8764.0ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.13 (= 43.8ms / 38.7ms)

OneFlow swin dataloader time: 0.259s (= 51.781s / 200, num_workers=1)
PyTorch swin dataloader time: 0.151s (= 30.118s / 200, num_workers=1)
Relative speed: 0.582 (= 0.151s / 0.259s)

OneFlow swin dataloader time: 0.066s (= 13.205s / 200, num_workers=4)
PyTorch swin dataloader time: 0.043s (= 8.598s / 200, num_workers=4)
Relative speed: 0.651 (= 0.043s / 0.066s)

OneFlow swin dataloader time: 0.038s (= 7.601s / 200, num_workers=8)
PyTorch swin dataloader time: 0.022s (= 4.482s / 200, num_workers=8)
Relative speed: 0.590 (= 0.022s / 0.038s)

❌ OneFlow resnet50 time: 145.2ms (= 14524.9ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 176.1ms (= 17607.9ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.21 (= 176.1ms / 145.2ms)

OneFlow resnet50 time: 97.3ms (= 9730.6ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 109.8ms (= 10979.0ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.13 (= 109.8ms / 97.3ms)

OneFlow resnet50 time: 76.7ms (= 15347.5ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 99.4ms (= 19870.2ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.29 (= 99.4ms / 76.7ms)

OneFlow resnet50 time: 65.3ms (= 13062.1ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 74.5ms (= 14908.1ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
❌ Relative speed: 1.14 (= 74.5ms / 65.3ms)

OneFlow resnet50 time: 55.1ms (= 11024.1ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 69.1ms (= 13816.6ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.25 (= 69.1ms / 55.1ms)