Arrow - nested types

[hjrnunes]

Seems TMD doesn't like Arrow nested types.

Consider the following example with List using Pandas:

import pandas as pd
import pyarrow.feather as ft

df_list = pd.DataFrame({'idx': [0, 1, 2],
                        'class-name': [['dog', 'car'], ['dog', 'flower'], ['car', 'flower']],
                        'confidence': [[0.8, 0.3], [0.75, 0.85], [0.46, 0.84]],
                        })

df_list.to_feather("arrow_list.arrow")

Loading arrow_list.arrow with TMD results in:

|                                                                           :$error |         :$value |
|-----------------------------------------------------------------------------------|-----------------|
| Failed to datafy datatype class org.apache.arrow.vector.types.pojo.ArrowType$List | arrow_img.arrow |

Extending Datafy to ArrowType$List further results in:

Execution error (AssertionError) at tech.v3.libs.arrow/records->ds (arrow.clj:1351).
Assert failed: (= (count fields) (count nodes))

So Field.children are not taken into account when reading the schema with tech.v3.libs.arrow/read-schema and the assertion fails because it'll then have, in this case, 5 nodes for 3 fields only.