Add explanations for all config options

**Is your feature request related to a problem? Please describe.**

There are over a 100 config options in the [standard llm recipes yaml files](https://github.com/NVIDIA-NeMo/RL/blob/main/examples/configs/recipes/llm/grpo-qwen2.5-32b-32n8g-fsdp2tp8sp-actckpt-long.v3.yaml) but they are not explained in the file or in the code. E.g. there are no explanations for `GRPOConfig`
```
class GRPOConfig(TypedDict):
num_prompts_per_step: int
num_generations_per_prompt: int
max_num_steps: int
max_rollout_turns: int
normalize_rewards: bool
use_leave_one_out_baseline: bool
val_period: int
val_batch_size: int
val_at_start: bool
max_val_samples: int
seed: int
```

**Describe the solution you'd like**

Explaining the config options in the code or having an example yaml file would help to understand some of the more nuanced options better. VERL does this pretty well: https://github.com/volcengine/verl/blob/main/verl/trainer/config/actor/actor.yaml.