`test_options_validator` fails with field mismatches (on Darwin)

[sarahec]

Happens while building docling-jobkit 1.1.1 and 1.2.0 for darwin (MacOS) in nixpkgs

tests/test_options.py:90: in test_options_validator
    assert (
E   AssertionError: assert InlineVlmOpti...w_tokens=4096) == InlineVlmOpti...w_tokens=4096)
E
E     Full diff:
E     - InlineVlmOptions(kind='inline_model_options', prompt='Convert this page to docling.', scale=2.0, max_size=None, repo_id='ds4sd/SmolDocling-256M-preview-mlx-bf16', trust_remote_code=False, load_in_8bit=True, llm_int8_threshold=6.0, quantized=False, inference_framework=<InferenceFramework.MLX: 'mlx'>, transformers_model_type=<TransformersModelType.AUTOMODEL: 'automodel'>, response_format=<ResponseFormat.DOCTAGS: 'doctags'>, torch_dtype=None, supported_devices=[<AcceleratorDevice.MPS: 'mps'>], temperature=0.0, stop_strings=[], extra_generation_config={}, use_kv_cache=True, max_new_tokens=4096)
E     ?                                                                                                                                                        ^^^^^^^^^^^                                                                                                                               ^^   ^^^
E     + InlineVlmOptions(kind='inline_model_options', prompt='Convert this page to docling.', scale=2.0, max_size=None, repo_id='ds4sd/SmolDocling-256M-preview', trust_remote_code=False, load_in_8bit=True, llm_int8
_threshold=6.0, quantized=False, inference_framework=<InferenceFramework.TRANSFORMERS: 'transformers'>, transformers_model_type=<TransformersModelType.AUTOMODEL_VISION2SEQ: 'automodel-vision2seq'>, response_format=<ResponseFormat.DOCTAGS: 'doctags'>, torch_dtype=None, supported_devices=[<AcceleratorDevice.CPU: 'cpu'>, <AcceleratorDevice.CUDA: 'cuda'>, <AcceleratorDevice.MPS: 'mps'>], temperature=0.0, stop_strings=[], extra_generation_config={}, use_kv_cache=True, max_new_tokens=4096)
E     ?                                                                                                                                                        ^^                                                                                                                              ++++++++ ^^^   ^^^^^^^^^^^^                                                            +++++++++++            +++++++++++                                                                                              ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++