[BUG] DataFrame.from_arrow results in incorrect order of child columns inside a StructColumn

[galipremsagar]

Describe the bug
Reading the same pyarrow table & array via two different APIs result in different results. Ideally both should have same correct results.
sample.parquet.zip

Steps/Code to reproduce bug

In [1]: import pyarrow as pa

In [2]: import pyarrow.parquet as pq

In [4]: table = pq.read_table('sample.parquet')

In [5]: import cudf

In [6]: cudf.DataFrame.from_arrow(table)['2']
Out[6]: 
0       {'0': None, '1': None, '2': None}
1         {'0': '', '1': None, '2': None}
2    {'0': None, '1': 'W&RR=+I', '2': ''}
Name: 2, dtype: struct

In [7]: cudf.Series.from_arrow(table['2'])
Out[7]: 
0       {'2': None, '0': None, '1': None}
1         {'2': '', '0': None, '1': None}
2    {'2': None, '0': 'W&RR=+I', '1': ''}
dtype: struct

Expected behavior

In [6]: cudf.DataFrame.from_arrow(table)['2']
Out[6]: 
0       {'2': None, '0': None, '1': None}
1         {'2': '', '0': None, '1': None}
2    {'2': None, '0': 'W&RR=+I', '1': ''}
Name: 2, dtype: struct

Environment overview (please complete the following information)

• Environment location: [Bare-metal]
• Method of cuDF install: [from source]