Fixed some bugs (see commit details and implementation for details)

[Syqinx]

已更正代码中的一些bug，请审查

[Syqinx]

@ShichangZh

[ShichangZh]

Thank you so much for bringing up these issues and providing potential solutions to help improve our code quality. I have a few comments.

1. For the output_path, it is indeed a bug when special parameters like feature_noise are set and num_exp > 1. Thank you for catching it. However, regarding the fix, I would prefer accessing num_exp inside the run function with args.num_exp instead of adding an extra argument to the run function. This will keep the function arguments clean.
2. For the implementation of the rand_train_test_idx function. I think there could be a mismatch between the intended non-negative node index and the actual non-negative node index when ignore_negative is set to true, as label = label[label >= 0] changes all the node indices without mapping them back.
3. What are these .idea/ files for?

Please kindly let me know whether you would agree with these comments.

[Syqinx]

1. Yes! You are right, but I think the parameter args.seed is a better choise, Because args.num_exp does not represent the current state of the experiment, it is just a number that specifies the number of times the experiment is performed.
2. Yes! Thank you for your correction!
3. .idea/ folder was added automatically when I imported the project into IDE-PyCharm, and I forgot to exclude it from the .gitignore file.

[Syqinx]

@ShichangZh

[ShichangZh]

Thank you for the quick update. I agree that using args.seed is a more reasonable choice for the first point. For the second point of the rand_train_test_idx function, a minor comment is that the variable name valid_idx is overloaded for "valid (nonnegative) labels" and "labels for validation". Maybe the first valid_idx in line 253 can be renamed to something like nonnegative_idx to avoid confusion?

[ShichangZh]

The fix looks good to me. I think this pull request can be merged. @nshah171