Onnxruntime

* Execution Provider updates
* MKL-DNN provider ([subgraph based execution](https://github.com/microsoft/onnxruntime/blob/master/docs/execution_providers/MKL-DNN-Subgraphs.md)) for improved performance
* Intel OpenVINO EP now available for Public Preview - [build instructions](https://github.com/microsoft/onnxruntime/blob/master/BUILD.md#openvino-build)
* Update to CUDA 10 for inferencing with NVIDIA GPUs
* Base CPU EP has faster convolution performance using the NCHWc blocked layout. This layout optimization can be enabled by setting graph optimization level to 3 in the session options.
* [C++ API](https://github.com/microsoft/onnxruntime/blob/master/include/onnxruntime/core/session/onnxruntime_cxx_api.h) for inferencing (wrapper on C API)
* [ONNX Runtime Server (Beta)](https://github.com/microsoft/onnxruntime/blob/master/docs/ONNX_Runtime_Server_Usage.md) for inferencing with HTTP and GRPC endpoints
* [Python Operator (Beta)](https://github.com/microsoft/onnxruntime/blob/master/docs/PyOp.md) to support custom Python code in a single node of an ONNX graph to make it easier for experimentation of custom operators
* Support of Keras-based Mask R-CNN model. The model relies on some custom operators pending to be added in ONNX; in the meantime, it can be converted using [this](https://github.com/onnx/keras-onnx/tree/master/applications/mask_rcnn) script for inferencing using ONNX Runtime 0.5. Other object detection models can be found from the [ONNX Model Zoo](https://github.com/onnx/models#object-detection--image-segmentation-).
* Minor updates to the C API
* For consistency, all C APIs now return an ORT status code
* Code coverage for this release is 83%

Key Updates
* New execution providers for improved performance on specialized hardware
* Intel nGraph
* NVIDIA TensorRT
* ONNX 1.5 compatibility
* Opset 10 operator support
* Supports newly added ONNX model zoo object detection models (YOLO v3, SSD)
* Quantization operators
* Updates to C API for Custom Operators
* Allocation of outputs during compute
* C++ wrapper to greatly simplify implementation
* Supports custom op DLLs when ONNX Runtime is compiled statically
* Graph optimizations with Constant Folding for improved performance
* Official binary packages
* Nuget package creation pipeline updated with security-focused tasks
* CredScan
* SDLNative Rules for PreFast
* BinSim
* Additional binaries built with MKL-ML published in Nuget
* Size reduction in Windows (700KB+), Linux (65%) and Mac (45%) binaries

This is a patch release for 0.3.0.

Updates include
* Binary size reduction through usage of protobuf-lite and operator fixes
* Build option to disable contrib ops (ops not in ONNX standard)
* Build option to statically link MSVC
* Minor bug fixes

Key Updates
* ONNX 1.4 compatibility
* Opset 9 operator support
* Support of large models >2GB

* New build packages
* C/C: OS X x64 CPU
* C: Linux x86 CPU
* C: Windows x86 CPU
* Custom op registration via C API
* Non-Tensor type support for input/output for C and C API

Release Notes
* Default execution provider for CPU uses Eigen and MLAS; prior releases used MKL-DNN. See all build options [here](https://github.com/Microsoft/onnxruntime/blob/rel-0.3.0/BUILD.md).
* OpenMP is required for the prebuilt binaries. See [System Requirements](https://github.com/Microsoft/onnxruntime/tree/rel-0.3.0#system-requirements) for more details.

Key Updates:
* ONNX Runtime C packages are now available for Linux, with GPU support for both Windows and Linux. Find the APIs and package downloads [here](https://github.com/Microsoft/onnxruntime/tree/rel-0.2.0#apis-and-official-builds).
* The [C API](https://github.com/Microsoft/onnxruntime/blob/rel-0.2.0/docs/C_API.md) has been updated and is now in Beta (previously: experimental). This version is expected to be mostly stable, though may adapt to ensure support of usage needs
* Support of additional operators with MKL-DNN: Relu, Sum, BatchNormalization

Release Notes
* The prebuilt-binaries in the CPU builds of the release require OpenMP at runtime. For Linux systems, it requires libgomp.so.1 to be installed. If OnnxRuntime fails to load, please try installing libgomp1.
* The binaries in the GPU builds require CUDA 9.1 and CuDNN 7.1 runtime libraries to be available in the system. For the Windows NuGet package of the v0.2.1 release, this is CUDA 10.0 and CuDNN 7.3 instead.

This is just a minor patch to the previous 0.1.4 release.