GPT2 torch.compile fix

[dsmertin]

What does this PR do?

Quick fix for this exception:

PT_HPU_LAZY_MODE=0 python run_generation.py --model_name_or_path gpt2 --batch_size 2 --max_new_tokens 100 --use_hpu_graphs --use_kv_cache --bf16 --bucket_size 50 --torch_compile

File "/root/optimum-habana/examples/text-generation/utils.py", line 187, in get_torch_compiled_model

    model.model = torch.compile(model.model, backend="hpu_backend", options={"keep_input_mutations": True})

  File "/usr/local/lib/python3.10/dist-packages/torch/nn/modules/module.py", line 1732, in __getattr__

    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")

AttributeError: 'GaudiGPT2LMHeadModel' object has no attribute 'model'

I'm also preparing better solution, but with more changes which can affect other models.

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?

[jiminha]

@regisss this also needs to pull in 1.14 release.