enlarge tolerance of dataloader test

[MARD1NO]

增大dataloader单测的tolerance

[github-actions]

Speed stats:

GPU Name: GeForce GTX 1080 

PyTorch resnet50 time: 140.7ms (= 7035.4ms / 50, input_shape=[16, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 127.0ms (= 6349.2ms / 50, input_shape=[16, 3, 224, 224], backward is enabled)
Relative speed: 1.11 (= 140.7ms / 127.0ms)

PyTorch resnet50 time: 84.3ms (= 4213.5ms / 50, input_shape=[8, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 74.1ms (= 3706.0ms / 50, input_shape=[8, 3, 224, 224], backward is enabled)
Relative speed: 1.14 (= 84.3ms / 74.1ms)

PyTorch resnet50 time: 58.9ms (= 2944.7ms / 50, input_shape=[4, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 49.5ms (= 2476.6ms / 50, input_shape=[4, 3, 224, 224], backward is enabled)
Relative speed: 1.19 (= 58.9ms / 49.5ms)

PyTorch resnet50 time: 49.8ms (= 2490.6ms / 50, input_shape=[2, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 42.8ms (= 2140.0ms / 50, input_shape=[2, 3, 224, 224], backward is enabled)
Relative speed: 1.16 (= 49.8ms / 42.8ms)

PyTorch resnet50 time: 44.4ms (= 2219.3ms / 50, input_shape=[1, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 42.2ms (= 2111.6ms / 50, input_shape=[1, 3, 224, 224], backward is enabled)
Relative speed: 1.05 (= 44.4ms / 42.2ms)