The split placement example is outdated, I tried it and encountered some
errors. To address this, the following changes were made in this PR
1. Copied the content from `verl/trainer/config/ppo_trainer.yaml` to
`examples/split_placement/config/ppo_trainer_split.yaml`
2. Copied `RayPPOTrainer.fit` method into the `fit` func in
`examples/split_placement/split_monkey_patch.py` and modified it to get
the futures of `critic_output` and `actor_output`
