feature: update wandb callback to upload checkpoints

[parambharat]

What does this PR do?

The PR updates the WandbCallback with the following changes:

• Adds on_save method to upload model checkpoints as artifacts.
• Changes the default value of environment variable WANDB_WATCH from gradients to false. This enables quicker training when defaults are used. The user can easily change this behavior by setting the env variable.
• Changes the WANDB_LOG_MODEL variable from bool to str allowing for different settings to upload artifacts.
• Modifies the class dostring to reflect the above changes.
• Fixes broken link to wandb documentation
• Changes the wandb run_name from output_dir to wandb auto generated name. this avoids duplication of run names in wandb workspace

Before submitting

• Did you read the contributor guideline,
  Pull Request section?
• Did you make sure to update the documentation with your changes? Here are the
  documentation guidelines, and
  here are tips on formatting docstrings.

Who can review?

• trainer: @sgugger
• documentation: @stevhliu

Examples

• Example colab reflecting all the changes to the WandbCallback
• Example Weights & Biases workspace with runs that show different settings.
• Example Weights & Biases Artifact created for checkpoints

[HuggingFaceDocBuilderDev]

The documentation is not available anymore as the PR was closed or merged.

[sgugger]

Thanks for your PR! I left a couple of comments.

[parambharat]

Hey, @sgugger : Thanks for the quick review and suggestions. I've resolved all the issues. 🤗

[sgugger]

@parambharat There is a weird issue with the tests. Could you try refreshing your permissions as shown here and then pushing an empty commit?

[parambharat]

Hey, @sgugger. I ran the fix that you suggested and all checks pass now.

[stevhliu]

Thanks for updating! I left two suggestions on formatting :)

[parambharat]

Thank you @stevhliu. I've committed your recommendations.

[sgugger]

Thanks again for your contribution!