BUG: using dtype='int64' argument of Series causes ValueError: values cannot be losslessly cast to int64 for integer strings

[5j9]

Pandas version checks

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• I have confirmed this bug exists on the master branch of pandas.

Reproducible Example

import pandas as pd
pd.Series(['1', '2'], dtype='int64')

Issue Description

C:\Users\a\AppData\Local\Programs\Python\Python310\lib\site-packages\numpy\core\numeric.py:2446: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison
  return bool(asarray(a1 == a2).all())
Traceback (most recent call last):
  File "C:\Users\a\AppData\Roaming\JetBrains\PyCharmCE2021.3\scratches\scratch_3.py", line 2, in <module>
    pd.Series(['1', '2'], dtype='int64')
  File "f:\prog\pandas\pandas\core\series.py", line 450, in __init__
    data = sanitize_array(data, index, dtype, copy)
  File "f:\prog\pandas\pandas\core\construction.py", line 583, in sanitize_array
    subarr = _try_cast(data, dtype, copy, raise_cast_failure)
  File "f:\prog\pandas\pandas\core\construction.py", line 768, in _try_cast
    subarr = maybe_cast_to_integer_array(arr, dtype)
  File "f:\prog\pandas\pandas\core\dtypes\cast.py", line 2100, in maybe_cast_to_integer_array
    raise ValueError(f"values cannot be losslessly cast to {dtype}")
ValueError: values cannot be losslessly cast to int64

Expected Behavior

I expect
pd.Series(['1', '2'], dtype='int64')
to work fine and return a Series object with dtype int64, just like how
pd.Series(['1', '2'], dtype='float64')
returns a float64 Series.

Installed Versions

INSTALLED VERSIONS

commit : 39ccb35
python : 3.10.1.final.0
python-bits : 64
OS : Windows
OS-release : 10
Version : 10.0.19043
machine : AMD64
processor : Intel64 Family 6 Model 94 Stepping 3, GenuineIntel
byteorder : little
LC_ALL : None
LANG : None
LOCALE : English_United States.1252
pandas : 0+untagged.1.g39ccb35
numpy : 1.21.4
pytz : 2021.3
dateutil : 2.8.2
pip : 21.3.1
setuptools : 59.5.0
Cython : 0.29.25
pytest : 6.2.5
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : 3.0.2
lxml.etree : 4.6.3
html5lib : 1.1
pymysql : None
psycopg2 : None
jinja2 : 3.0.3
IPython : 7.29.0
pandas_datareader: None
bs4 : 4.10.0
bottleneck : None
fsspec : None
fastparquet : None
gcsfs : None
matplotlib : 3.5.0
numexpr : None
odfpy : None
openpyxl : 3.0.9
pandas_gbq : None
pyarrow : None
pyxlsb : None
s3fs : None
scipy : None
sqlalchemy : None
tables : None
tabulate : None
xarray : None
xlrd : None
xlwt : None
numba : None

[simonjayhawkins]

Expected Behavior

I expect pd.Series(['1', '2'], dtype='int64') to work fine and return a Series object with dtype int64

This was the result in 1.2.5

first bad commit: [db6e71b] DEPR: silent overflow on Series construction (#41734)

cc @jbrockmendel

also note that we have a rogue /home/simon/miniconda3/envs/pandas-1.2.5/lib/python3.9/site-packages/numpy/core/numeric.py:2446: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison return bool(asarray(a1 == a2).all()) bubbling up from numpy from that looks like a more long-standing issue. (at least 0.25.3)

[jbrockmendel]

Looks like in maybe_cast_to_integer_array the check for np.array_equal(arr, casted) isn't the appropriate check in this case. might be we need to cast back to an original dtype before doing the array_equal check