Bourse

Add simulation environment that allow instructions to be submitted and returned directly in a Numpy array format.

Various fixes to documentation

- Add built in momentum agent type
- Track level 2 data during simulation

- Add pre-built random and noise agent implementations
- Switch random number generation from `fastrand` to generic rust [rand crates](https://rust-random.github.io/book/)
- Implement serde serialize and deserialize traits
- Add the ability in Rust and Python to export and load order book state from JSON files

Order book now tracks number of individual orders placed at price levels. The number of orders at the touch is then tracked by the simulation environment.

- Verify Python documentation examples with doctest
- More and more detailed Rust examples
- Update documentation links