Fix updates in random forest model using GOSS data sample strategy

[mjmckp]

1) Value of data_sample_strategy was not written out in Config::SaveMembersToString()
2) GOSSStrategy->Bagging may modify value of bag_data_cnt_ during training, which may mean tmp_grad_ and tmp_hess_ need resizing in RF::TrainOneIter. Without this change, an exception is thrown during training at line 138 of rf.hpp.

[mjmckp]

@jameslamb Are we ready to merge this now please?

[jameslamb]

Are we ready to merge this now please?

I've just merged master into this, and will merge it if/when it builds successfully.

In the future, you can reduce the time it takes to get changes reviewed and merged here by doing the following in pull requests:

• clearly describing why changes are being made, not just what you're changing (ideally with a minimal reproducible example and/or new tests demonstrating the change in behavior)
• linking directly to prior conversations and/or any relevant code, to help reviewers get context on exactly why you're proposing a set of changes
• putting up separate PRs for unrelated concerns (in this example, updates in random forest model and changes to the model file could have been separated)

Thanks as always for your help with LightGBM!

[github-actions]

This pull request has been automatically locked since there has not been any recent activity since it was closed. To start a new related discussion, open a new issue at https://github.com/microsoft/LightGBM/issues including a reference to this.