[Feature Request] Enable CUDA Execution Provider in Rust binding

### Describe the feature request

Enable the CUDA execution provider through the Rust binding.

This seems to be supported by [`onnxruntime-sys`](https://github.com/microsoft/onnxruntime/blob/main/rust/onnxruntime-sys/examples/c_api_sample.rs#L68-L70), but it doesn't appear to be exposed by the Rust binding. I believe it is relatively simple, by exposing the `OrtSessionOptionsAppendExecutionProvider_CUDA` function. If a PR would be accepted, let me know :).

### Describe scenario use case

Using CUDA as an execution provider.

Start by inspecting the Rust binding and the onnxruntime-sys C API example linked in the issue, especially the OrtSessionOptionsAppendExecutionProvider_CUDA reference. Compare the exposed execution-provider functions with that example. Done means the CUDA execution provider is available through the Rust binding and the existing binding checks pass.

Written by the indexing model from the issue text.