avoid extra reshaping to max_model_lenght for unet

[eaidova]

What does this PR do?

unet in stable diffusion pipelines always is saved with dynamic shapes during export and uses reshape to set max_model_lenght from tokenizer before starting inference. Suggestion is remove reshape every model loading and set max_model_len as default value for encoder_hidden_state during export if it is known

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you make sure to update the documentation with your changes?
• Did you write any new necessary tests?

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

[echarlaix]

LGTM!