[Rust] GrpoRewards: Implement Utility Functions

[kyjohnso]

@nikg4 @taenin what other functions should we put into the GrpoRewards struct? other rewards functions? Here is a first cut from sonnet

4. Utility Functions:
   • parallel_reward_computation(trajectories) - Leverage Rust's performance for batch processing
   • serialize_reward_function(reward_fn) - Save/load custom reward functions
   • vectorize_reward(states, actions) - Efficient batch reward computation

Implement parallel_reward_computation, serialize_reward_function, vectorize_reward

Originally posted by @kyjohnso in #13