refine error msg: add NOLINT for defensive code in autograd

[pingzhuu]

• 完善 autograd 中防御代码的错误信息提示，添加 // NOLINT(maybe-need-error-msg)
• 涉及的文件比较多，所以将只涉及几种简单情况的单独这个 PR 中，便于 review
• 这个 PR 中包含的文件只处理以下几种类型，均位于 Init, Capture, Apply 函数开头检查参数数量是否正确
  • CHECK_NOTNULL_OR_RETURN(fw_op_expr);
  • CHECK_EQ_OR_RETURN(inputs.size(), xxx);
  • CHECK_EQ_OR_RETURN(outputs.size(), xxx);
  • CHECK_EQ_OR_RETURN(out_grads.size(), xxx);

[hjchen2]

注意一下，需要将这些被改动过的文件中所有报错信息缺失的地方都改过来，才能通过need-check-error-message的检查

[pingzhuu]

注意一下，需要将这些被改动过的文件中所有报错信息缺失的地方都改过来，才能通过need-check-error-message的检查

嗯，这个确认过，改动过的文件有 CHECK 的都改过了

[github-actions]

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/8525/

[github-actions]

Speed stats:

GPU Name: NVIDIA GeForce GTX 1080 

❌ OneFlow resnet50 time: 129.1ms (= 12906.2ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 144.6ms (= 14458.7ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.12 (= 144.6ms / 129.1ms)

OneFlow resnet50 time: 77.1ms (= 7710.3ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 86.0ms (= 8604.5ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.12 (= 86.0ms / 77.1ms)

OneFlow resnet50 time: 49.7ms (= 9936.8ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 58.8ms (= 11750.2ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.18 (= 58.8ms / 49.7ms)

OneFlow resnet50 time: 41.0ms (= 8202.7ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 43.5ms (= 8695.9ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.06 (= 43.5ms / 41.0ms)

OneFlow resnet50 time: 36.4ms (= 7289.1ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 39.6ms (= 7916.4ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.09 (= 39.6ms / 36.4ms)

OneFlow swin dataloader time: 0.270s (= 54.057s / 200, num_workers=1)
PyTorch swin dataloader time: 0.152s (= 30.409s / 200, num_workers=1)
Relative speed: 0.563 (= 0.152s / 0.270s)

OneFlow swin dataloader time: 0.081s (= 16.234s / 200, num_workers=4)
PyTorch swin dataloader time: 0.041s (= 8.243s / 200, num_workers=4)
Relative speed: 0.508 (= 0.041s / 0.081s)

OneFlow swin dataloader time: 0.045s (= 8.909s / 200, num_workers=8)
PyTorch swin dataloader time: 0.022s (= 4.421s / 200, num_workers=8)
Relative speed: 0.496 (= 0.022s / 0.045s)

❌ OneFlow resnet50 time: 145.0ms (= 14503.1ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 171.3ms (= 17132.4ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.18 (= 171.3ms / 145.0ms)

OneFlow resnet50 time: 93.3ms (= 9333.9ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 113.2ms (= 11324.7ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.21 (= 113.2ms / 93.3ms)

OneFlow resnet50 time: 70.3ms (= 14065.0ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 91.3ms (= 18264.2ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.30 (= 91.3ms / 70.3ms)

OneFlow resnet50 time: 57.6ms (= 11521.0ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 76.2ms (= 15244.3ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.32 (= 76.2ms / 57.6ms)

OneFlow resnet50 time: 52.7ms (= 10543.7ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 69.2ms (= 13834.2ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.31 (= 69.2ms / 52.7ms)