Expose a rust lib in addition to the python lib

[peterschutt]

I've now got a version of this library that can also be used in rust/cargo projects. You can cargo add it, and use it like this:

let odds = [2.0, 3.5, 4.0];

// Simple usage
let result = implied_probabilities(&odds)
    .solve()
    .unwrap();
let probabilities = &result.probs;

// With custom parameters
let result = implied_probabilities(&odds)
    .with_max_iterations(1000)
    .with_convergence_threshold(1e-12)
    .solve()
    .unwrap();

// Accessing full details
println!("Probabilities: {:?}", result.probs);
println!("Iterations: {}", result.iterations);
println!("Delta: {}", result.delta);
println!("Z value: {}", result.z);

To do this, I've pushed all of the logic of calculate_implied_probabilities() except for the odds mapping input conversion logic down into the rust implementation. So the python now looks like this:

...
from .shin import calculate_implied_probabilities as _calculate_implied_probabilities

def calculate_implied_probabilities(
    odds: Sequence[float] | Mapping[T, float],
    *,
    max_iterations: int = 1000,
    convergence_threshold: float = 1e-12,
    full_output: bool = False,
) -> (
    ShinOptimisationDetails[list[float]]
    | ShinOptimisationDetails[dict[T, float]]
    | list[float]
    | dict[T, float]
):
    odds_seq = list(odds.values()) if isinstance(odds, Mapping) else odds

    res = _calculate_implied_probabilities(
        odds_seq,
        max_iterations=max_iterations,
        convergence_threshold=convergence_threshold,
    )

    if isinstance(odds, Mapping):
        d = dict(zip(odds, res.probs, strict=False))
        if full_output:
            return ShinOptimisationDetails(
                implied_probabilities=d,
                iterations=res.iterations,
                delta=res.delta,
                z=res.z,
            )
        return d

    if full_output:
        return ShinOptimisationDetails(
            implied_probabilities=res.probs,
            iterations=res.iterations,
            delta=res.delta,
            z=res.z,
        )
    return res.probs

Note that I've removed the pure python optimizer functionality, but that could still be retained if you'd like.

Any interest in making similar changes here?