Restruct MathOps AutoTest

[BBuf]

把所有的三角函数替换为新版自动测试框架，删除之前的手动测试。

所有三角函数的单独测试文件均删除，移动到math_ops的测试文件中。

另外修复了自动测试框架制定Tensor的数值范围失效的BUG，新增了避免死循环的循环次数检查。

新增了一些其它自动测试例子，Conv2d，Floor等等。

[MARD1NO]

out channels这里要不要设置小点，跑起来快点？

[BBuf]

好的

[MARD1NO]

和833行测试重复了，删除下

[BBuf]

好的

[github-actions]

Speed stats:

GPU Name: GeForce GTX 1080 

PyTorch resnet50 time: 140.6ms (= 7030.7ms / 50, input_shape=[16, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 126.2ms (= 6309.0ms / 50, input_shape=[16, 3, 224, 224], backward is enabled)
Relative speed: 1.11 (= 140.6ms / 126.2ms)

PyTorch resnet50 time: 83.7ms (= 4185.1ms / 50, input_shape=[8, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 73.7ms (= 3686.7ms / 50, input_shape=[8, 3, 224, 224], backward is enabled)
Relative speed: 1.14 (= 83.7ms / 73.7ms)

PyTorch resnet50 time: 57.2ms (= 2861.9ms / 50, input_shape=[4, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 47.7ms (= 2383.1ms / 50, input_shape=[4, 3, 224, 224], backward is enabled)
Relative speed: 1.20 (= 57.2ms / 47.7ms)

PyTorch resnet50 time: 49.1ms (= 2456.9ms / 50, input_shape=[2, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 41.7ms (= 2087.3ms / 50, input_shape=[2, 3, 224, 224], backward is enabled)
Relative speed: 1.18 (= 49.1ms / 41.7ms)

PyTorch resnet50 time: 43.6ms (= 2181.0ms / 50, input_shape=[1, 3, 224, 224], backward is enabled)
OneFlow resnet50 time: 48.2ms (= 2410.9ms / 50, input_shape=[1, 3, 224, 224], backward is enabled)
Relative speed: 0.90 (= 43.6ms / 48.2ms)