[BUG] Max with null in datetime columns gives incorrect results

[VibhuJawa]

Describe the bug
Max with null in one of the datetime columns gives nulls instead of actual results

Steps/Code to reproduce bug

import cudf
import pandas as pd
from datetime import datetime

p_df = pd.DataFrame({
    "datetime_0": [datetime(2011, 4, 9, 10, 30, 15)]*4,
    "datetime_1": [None]*4})
p_df['datetime_1']=p_df['datetime_1'].astype('datetime64[ns]')

df = cudf.from_pandas(p_df)
print(df.max(axis=1))

0    <NA>
1    <NA>
2    <NA>
3    <NA>
dtype: datetime64[ns]

Expected behavior

import cudf
import pandas as pd
from datetime import datetime

p_df = pd.DataFrame({
    "datetime_0": [datetime(2011, 4, 9, 10, 30, 15)]*4,
    "datetime_1": [None]*4})
p_df['datetime_1']=p_df['datetime_1'].astype('datetime64[ns]')

print(p_df.max(axis=1))

0   2011-04-09 10:30:15
1   2011-04-09 10:30:15
2   2011-04-09 10:30:15
3   2011-04-09 10:30:15
dtype: datetime64[ns]

Environment overview (please complete the following information)

• Method of cuDF install: [conda]

cudf                      0.17.0a201209   cuda_10.2_py37_gbd321d1e93_382    rapidsai-nightly
libcudf                   0.17.0a201209   cuda10.2_gbd321d1e93_382    rapidsai-nightly