TypeError: read_json() got an unexpected keyword argument 'header'

```
ModelError: Received server error (500) from primary with message "read_json() got an unexpected keyword argument 'header'
Traceback (most recent call last):
File "/opt/conda/lib/python3.11/site-packages/sagemaker_inference/transformer.py", line 150, in transform
result = self._run_handler_function(
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/opt/conda/lib/python3.11/site-packages/sagemaker_inference/transformer.py", line 280, in _run_handler_function
result = func(*argv)
^^^^^^^^^^^
File "/opt/ml/model/code/tabular_serve.py", line 55, in transform_fn
data = _read_with_fallback(pd.read_json, buf, column_names)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/opt/ml/model/code/tabular_serve.py", line 141, in _read_with_fallback
data = read_func(buf, header=None)
^^^^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: read_json() got an unexpected keyword argument 'header'
```

Exact place where issue occurs: https://github.com/autogluon/autogluon-cloud/blob/f515ec202fad7cc1ca3fd5687335cdd6ca2f3ae5/src/autogluon/cloud/scripts/sagemaker_scripts/tabular_serve.py#L141

Afaik, `header` argument is only available in `read_csv`. Will try to fix this if this will become a blocker.

The failure is at line 141 of src/autogluon/cloud/scripts/sagemaker_scripts/tabular_serve.py, inside _read_with_fallback. Inspect how pandas read_json and read_csv are called there, then reproduce the tabular serving path with JSON input. Done means the SageMaker request no longer raises the unexpected keyword argument error.

Written by the indexing model from the issue text.