Applying groupby and aggregating with nunique over float column with nulls unexpectedly changes dataframe

[cuckookernel]

Code Sample, a copy-pastable example if possible

import pandas as pd
import numpy as np
print( pd.__version__ )  # 1.0.1
df = pd.DataFrame( [[1, 0], [2, np.nan], [3, np.nan]], columns=['a', 'b'] )
print( df )
#   a    b
# 0  1  0.0
# 1  2  NaN
# 2  3  NaN
unused = df.groupby('a').agg({'b': 'nunique'})     # this shouldn't change anything in df, right?
print( df ) # prints out
#   a             b
# 0  1  0.000000e+00
# 1  2 -9.223372e+18
# 2  3 -9.223372e+18
# what happened to my nans?

Problem description

For some reason the agg, groupby operation is changing the orginal nan values in the b column
to a special float value.
In the previous version of pandas I was using, this didn't happen.

Expected Output

The expected output of the print should the same as the one of the first print as groupby + agg should never change the input df (right?)

Output of pd.show_versions()

[paste the output of ``pd.show_versions()`` here below this line]

INSTALLED VERSIONS

commit : None
python : 3.7.6.final.0
python-bits : 64
OS : Linux
OS-release : 4.15.0-70-generic
machine : x86_64
processor :
byteorder : little
LC_ALL : None
LANG : sv_SE.UTF-8
LOCALE : sv_SE.UTF-8
pandas : 1.0.1
numpy : 1.18.1
pytz : 2019.3
dateutil : 2.8.1
pip : 20.0.2
setuptools : 45.2.0.post20200210
Cython : None
pytest : None
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : None
html5lib : None
pymysql : None
psycopg2 : 2.8.4 (dt dec pq3 ext lo64)
jinja2 : 2.11.1
IPython : None
pandas_datareader: None
bs4 : None
bottleneck : None
fastparquet : None
gcsfs : None
lxml.etree : None
matplotlib : None
numexpr : None
odfpy : None
openpyxl : None
pandas_gbq : None
pyarrow : None
pytables : None
pytest : None
pyxlsb : 1.0.6
s3fs : None
scipy : 1.4.1
sqlalchemy : 1.3.13
tables : None
tabulate : None
xarray : None
xlrd : None
xlwt : None
xlsxwriter : None
numba : None

[MarcoGorelli]

Thanks @cuckookernel for the report!

This is indeed a bug - closing though as we have it reported in #31950