Module prelude 

Prelude module for convenient imports.

Brings the highest-value public types into scope with a single use thrust_rl::prelude::*;. The set is curated, not a blanket glob:

• Environment API (always available): the Environment trait plus StepResult, SpaceInfo, SpaceType, and the built-in CartPole / SimpleBandit environments.
• Training stack (requires the training feature): the MlpBurnPolicy actor-critic network and its MlpBurnConfig, the trainer configs/trainers (A2C, PPO, DQN, SAC, BC), and the RolloutBuffer / ReplayBuffer experience buffers, and the EnvPool vectorized env wrapper.

Training-only re-exports are #[cfg(feature = "training")]-gated so --no-default-features still compiles.

Re-exports

pub use crate::buffer::replay::ReplayBuffer;

pub use crate::buffer::rollout::RolloutBuffer;

pub use crate::env::pool::EnvPool;

pub use crate::env::CartPole;

pub use crate::env::Environment;

pub use crate::env::SimpleBandit;

pub use crate::env::SpaceInfo;

pub use crate::env::SpaceType;

pub use crate::env::StepResult;

pub use crate::policy::mlp::BurnActivation;

pub use crate::policy::mlp::MlpBurnConfig;

pub use crate::policy::mlp::MlpBurnPolicy;

pub use crate::train::A2cConfig;

pub use crate::train::A2cTrainer;

pub use crate::train::BcConfig;

pub use crate::train::BcTrainer;

pub use crate::train::DQNConfig;

pub use crate::train::DQNTrainerBurn;

pub use crate::train::PPOConfig;

pub use crate::train::PPOTrainerBurn;

pub use crate::train::SacConfig;

pub use crate::train::SacTrainer;