feat: add gpu mem and util logging to wandb/tensorboard

[terrykong]

What does this PR do ?

• Adds a simple unified way to log gpu memory and utilization to tensorboard and wandb
• Move the logger initialization earlier in each algorithm setup() so we can log memory/util ealier (previously the logging started after vLLM workers were spun up)
• small fix to repr since I observed it's possible to fail if torch.distributed hadn't been setup yet
• turn sft wandb logging default to False so it matches grpo and doesn't error if someone doesn't set WANDB_API_KEY
• vllm tests now work on a slurm ray cluster. previously they checked locally for gpus, which won't work since our head node has no GPUs. can revisit this when we look at mocking ray

Issues

List issues that this PR closes (syntax):

Closes #35

Usage

• You can potentially add a usage example below

# Add a code snippet demonstrating how to use this 

Before your PR is "Ready for review"

Pre checks:

• Make sure you read and followed Contributor guidelines
• Did you write any new necessary tests?
• Did you run the unit tests and functional tests locally? Visit our Testing Guide for how to run tests
• Did you add or update any necessary documentation? Visit our Document Development Guide for how to write, build and test the docs.

Additional Information

• ...

[terrykong]

Functional tests need a followup PR to fix #41

Running them manually, will merge once they pass