Support for microsoft / Phi-3.5-vision-instruct

[Youho99]

In the same idea of ​​support as phi-3-vision, and Florence-2 (see #658), integrate microsoft/Phi-3.5-vision-instruct which has just been released, in onnx in onnxruntime-genai.

There is a chance that the integration will be similar to that already done with phi-3-vision.

Thanks :)

[Elephant Panda (elephantpanda)]

Yes, please. But also Phi 3 vision int4 (DirectML) isn't quite working yet so please don't forget about that one unless phi 3.5 vision solves those problems.

[kunal-vaishnavi]

The work is in progress to support Phi-3.5 vision. For example, this PR will add support for multiple images. We are also looking into the DirectML issues with Phi-3 vision.

[kunal-vaishnavi]

The new Phi-3 vision and Phi-3.5 vision ONNX models have now been released. The new models support no-image, single-image, and multi-image scenarios.