[FEA] Support list types in read_json

[randerzander]

test.json:

{"id": 0, "val": [{"sub_id":"a"},{"sub_id": "b"}]}
{"id": 1, "val": [{"sub_id":"c"},{"sub_id":"c"}]}

In Pandas, val becomes a list of strings:

import pandas as pd
fn = 'test.json'

pd.read_json(fn, lines=True)
   id                                 val
0   0  [{'sub_id': 'a'}, {'sub_id': 'b'}]
1   1  [{'sub_id': 'c'}, {'sub_id': 'c'}]

In cudf, the file will load, but the list of JSON objects is parsed incorrectly.

import cudf
fn = 'test.json'
cudf.read_json(fn, lines=True)
   id              val sub_id
0   0  [{"sub_id":"a"}   b"}]
1   1  [{"sub_id":"c"}   c"}]

[vuule]

closing as #8827 is filed for the same feature