Fix tensor deepcopy bug

[lixiang007666]

This PR is done:

• Modify test_tensor_deepcopy func.
• Fix the bug of deepcopy of tensor, related issue: [BUG]Incorrect device of oneflow tensor after deepcopy #7428
• Open the Graph test for loss and tensor_split.
• Remove the TODO in GetDualObject.

[strint]

deepcopy的协议：

def _deepcopy_inst(x, memo):
    if hasattr(x, '__deepcopy__'):
        return x.__deepcopy__(memo)
    if hasattr(x, '__getinitargs__'):
        args = x.__getinitargs__()
        args = deepcopy(args, memo)
        y = x.__class__(*args)
    else:
        y = _EmptyClass()
        y.__class__ = x.__class__
    memo[id(x)] = y
    if hasattr(x, '__getstate__'):
        state = x.__getstate__()
    else:
        state = x.__dict__
    state = deepcopy(state, memo)
    if hasattr(y, '__setstate__'):
        y.__setstate__(state)
    else:
        y.__dict__.update(state)
    return y

Ref:
copy实现：https://svn.python.org/projects/python/trunk/Lib/copy.py
stackoverflow答案：https://stackoverflow.com/questions/22388854/relationship-between-pickle-and-deepcopy

[github-actions]

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

[github-actions]

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

[github-actions]

Speed stats:

GPU Name: GeForce GTX 1080 

✔️ OneFlow resnet50 time: 128.7ms (= 12868.5ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 142.1ms (= 14214.8ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.10 (= 142.1ms / 128.7ms)

✔️ OneFlow resnet50 time: 78.6ms (= 7860.8ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 84.9ms (= 8486.4ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.08 (= 84.9ms / 78.6ms)

OneFlow resnet50 time: 52.3ms (= 10464.4ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 57.9ms (= 11578.8ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.11 (= 57.9ms / 52.3ms)

OneFlow resnet50 time: 44.6ms (= 8922.7ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 48.5ms (= 9693.0ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.09 (= 48.5ms / 44.6ms)

OneFlow resnet50 time: 39.4ms (= 7872.3ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 39.4ms (= 7885.2ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.00 (= 39.4ms / 39.4ms)

✔️ OneFlow resnet50 time: 140.5ms (= 14053.9ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 158.8ms (= 15876.7ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.13 (= 158.8ms / 140.5ms)

OneFlow resnet50 time: 89.2ms (= 8921.1ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 103.0ms (= 10300.6ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.15 (= 103.0ms / 89.2ms)

OneFlow resnet50 time: 63.3ms (= 12665.3ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 78.0ms (= 15594.9ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.23 (= 78.0ms / 63.3ms)

OneFlow resnet50 time: 53.3ms (= 10656.4ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 63.2ms (= 12635.9ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.19 (= 63.2ms / 53.3ms)

OneFlow resnet50 time: 47.0ms (= 9392.4ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 58.8ms (= 11756.8ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.25 (= 58.8ms / 47.0ms)