[Feature Request] Implement encoder_hidden_states as input in GPT2_BeamSearch Node

### Describe the feature request

I try to use the [convert_generation.py](https://github.com/microsoft/onnxruntime/blob/main/onnxruntime/python/tools/transformers/convert_generation.py) script to create a GPT2 code generation model with beam search with encoder_hidden_states (timesformer output) as input (my base model is [Neleac/timesformer-gpt2-video-captioning](https://huggingface.co/Neleac/timesformer-gpt2-video-captioning)), but there's no such flags in scripts or node input in graph. So GPT2 coverting as separate model without link to timesformer output.

So I was wondering if there are any plans to implement this option. I've tried manually manipulating the graph and script to no avail.

### Describe scenario use case

Usage of Encoder-Decoder (such as SpeechEncoderDecoderModel or VisionEncoderDecoderModel from HF)

Start by reading onnxruntime/python/tools/transformers/convert_generation.py and locating the GPT2_BeamSearch Node definition and graph inputs it creates. Compare the current GPT-2 beam-search path with the encoder-decoder scenario described in the issue, then identify the relevant tests or model-conversion checks. Done means the conversion script accepts encoder_hidden_states and the generated graph exposes and uses it for the requested model.

Written by the indexing model from the issue text.