Add Inferentai2 and Optimum Neuron Support

[philschmid]

What does this PR do?

This PR add support for infernetia2. To deploy a model on Inferentia2 you have 3 options:

• Provide an already compiled model with a model.neuron file as HF_MODEL_ID, .e.g. optimum/tiny_random_bert_neuron
• Provide the HF_OPTIMUM_BATCH_SIZE and HF_OPTIMUM_SEQUENCE_LENGTH environment variables to compile the model on the fly, e.g. HF_OPTIMUM_BATCH_SIZE=1 HF_OPTIMUM_SEQUENCE_LENGTH=128
• Include neuron dictionary in the config.json file in the model archive, e.g. neuron: {"static_batch_size": 1, "static_sequence_length": 128}

The currently supported tasks can be found here. If you plan to deploy an LLM, we recommend taking a look at Neuronx TGI, which is purposly build for LLMs

[JingyaHuang]

LGTM, thanks @philschmid. Just left some small nits.