Decompose ndarray reduce in out dtype and dev all any

[simonJJJ]

Successfully get correct tensor_str and results.

[github-actions]

CI failed when running job: cpu-legacy-op. PR label automerge has been removed

[github-actions]

CI failed when running job: cpu-legacy-op. PR label automerge has been removed

[github-actions]

CI failed when running job: cpu-legacy-op. PR label automerge has been removed

[github-actions]

Speed stats:

GPU Name: GeForce GTX 1080 

OneFlow resnet50 time: 136.9ms (= 13694.0ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 143.2ms (= 14320.2ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.05 (= 143.2ms / 136.9ms)

OneFlow resnet50 time: 78.6ms (= 7856.2ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 86.4ms (= 8640.4ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.10 (= 86.4ms / 78.6ms)

OneFlow resnet50 time: 52.4ms (= 10480.3ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 59.0ms (= 11806.1ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.13 (= 59.0ms / 52.4ms)

OneFlow resnet50 time: 41.2ms (= 8241.0ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 46.8ms (= 9361.2ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.14 (= 46.8ms / 41.2ms)

OneFlow resnet50 time: 38.7ms (= 7744.0ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 39.0ms (= 7791.0ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.01 (= 39.0ms / 38.7ms)

OneFlow resnet50 time: 156.5ms (= 15646.5ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 161.5ms (= 16145.9ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.03 (= 161.5ms / 156.5ms)

OneFlow resnet50 time: 98.5ms (= 9850.6ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 105.0ms (= 10499.6ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.07 (= 105.0ms / 98.5ms)

OneFlow resnet50 time: 74.6ms (= 14925.2ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 77.5ms (= 15492.5ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.04 (= 77.5ms / 74.6ms)

OneFlow resnet50 time: 66.0ms (= 13204.8ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 68.5ms (= 13691.0ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.04 (= 68.5ms / 66.0ms)

OneFlow resnet50 time: 66.2ms (= 13246.2ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 58.2ms (= 11631.7ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 0.88 (= 58.2ms / 66.2ms)