would it be possible to use nbdime/ndbiff inside PySpark Notebooks? Often these type of notebooks have a specialized jupyter notebook format, e.g. Azure Synapse Analytics Notebooks.