ort2 0.1.1

onnxruntime wrapper c/c++ api
Documentation
# ort2
- onnxruntime wrapper for rust 2
- tested


| OS     | Linux | Windows | MacOS      |
|--------|-------|---------|------------|
| CPU    | Y     | Y       | Y(aarch64) |
| CUDA   | Y     | Y       | N/A        |


## Pre-requirements

- download onnxruntime from onnxruntime github [Repo]https://github.com/microsoft/onnxruntime/releases, unzip it
- setup enviroment variable
    - `ORT_INC_PATH=/opt/homebrew/opt/onnxruntime/include`
    - `ORT_LIB_PATH=/opt/homebrew/opt/onnxruntime/lib`

## Getting Started

```rust
use ort2::prelude::*;

// load model stuff
let model = include_bytes!("models/mnist-8.onnx");

// create session
let session = Session::builder()
    .build(model.as_ref())
    .expect("failed to create session");

// dump input
let input = vec![0.0f32;28 * 28];

// create value from input
let value = Value::tensor()
    .with_shape([1, 1, 28, 28])
    .with_typ(ONNXTensorElementDataType::ONNX_TENSOR_ELEMENT_DATA_TYPE_FLOAT)
    .borrow(&input)
    .expect("failed to build value");

// get output
let output = session.run([&value])
    .expect("failed to run")
    .into_iter()
    .next()
    .expect("failed to get outputs");

// output of session as ndarray Array
let output = output
    .view::<f32>()
    .expect("failed to view output");

assert_eq!(output.shape()[1], 10);
```