Add onnx support for VisionEncoderDecoder (#19254)

* Add onnx support for VisionEncoderDecoder

* Add onnx support for VisionEncoderDecoder

* Removed unused import

* Rename encoder hidden state

Co-authored-by: lewtun <lewis.c.tunstall@gmail.com>

* Update docstrings and removed redundant code

* Added test function for enc-dec models

* Update doc string text

Co-authored-by: lewtun <lewis.c.tunstall@gmail.com>

* fixed code style

Co-authored-by: lewtun <lewis.c.tunstall@gmail.com>

docs/source/en/serialization.mdx

@@ -96,6 +96,7 @@ Ready-made configurations include the following architectures:
 - SqueezeBERT
 - Swin Transformer
 - T5
+- Vision Encoder decoder
 - ViT
 - XLM
 - XLM-RoBERTa
@@ -294,6 +295,13 @@ that can be used for fast autoregressive decoding.
 
 </Tip>
 
+<Tip>
+
+For `VisionEncoderDecoder` type models, the encoder and decoder parts are
+exported separately as two ONNX files named `encoder_model.onnx` and `decoder_model.onnx` respectively.
+
+</Tip>
+
 
 ## Exporting a model for an unsupported architecture