[Feature Request] Is there any way to use OpenVINO EP and TensorRT EP in the same environment?

### Describe the feature request

I'm using python, I know if I want to use TensorRT EP, I have to use onnxruntime-gpu. If I want to use OPENVINO I have to use onnxruntime-openvino. But These cannot be installed in the same docker container. They overlap each other. I want to use both while don't want to build from source.

### Describe scenario use case

We use both CPU and GPU for inference of a model. And we want to use a single docker image.

The issue names no files, tests, or entry points. Start by reviewing how the onnxruntime-gpu and onnxruntime-openvino Python packages are built and installed in a Docker container; done would be a documented supported way to use both execution providers without building from source.

Written by the indexing model from the issue text.