Allow JSON encoder to handle ndarray

[srowen]

Description of changes:

The JSON encoder encodes a variety of data types via json.dumps, including primitives, lists, dicts, etc.
However json.dumps doesn't work with ndarrays ; it won't serialize them.
Conceptually, there isn't a strong reason to refuse to encode input as JSON where ndarray is used instead of lists, when one could simply transform even multi-dimensional ndarrays to supported Python lists with .tolist().

This change simply has the encoder call .tolist() on its arg if passed an ndarray, and the rest works as expected.

This is actually relevant because the serialization of array type data from Spark to pandas will use ndarrays (being based on Arrow), and the result can't be passed to dataframe_to_mds even if the output type is given as 'json'. It's possible to workaround by passing a transformation of the pandas DF to this function. But, it seemed simple and natural to just support this directly.

Merge Checklist:

General

• I have read the contributor guidelines
• This is a documentation change or typo fix. If so, skip the rest of this checklist.
• I certify that the changes I am introducing will be backward compatible, and I have discussed concerns about this, if any, with the MosaicML team.
• I have updated any necessary documentation, including README and API docs (if appropriate).

Tests

• I ran pre-commit on my change. (check out the pre-commit section of prerequisites)
• I have added tests that prove my fix is effective or that my feature works (if appropriate).
• I ran the tests locally to make sure it pass. (check out testing)
• I have added unit and/or integration tests as appropriate to ensure backward compatibility of the changes.

[snarayan21]

why not just use the ndarray encoding type?

[snarayan21]

Lgtm