[SPARK-52771][PS] Fix float32 type widening in truediv/floordiv

[xinrong-meng]

What changes were proposed in this pull request?

Fix float32 type widening in truediv/floordiv when ANSI on/off.

Why are the changes needed?

Ensure pandas on Spark works well with ANSI mode on/off.

Note that the issue exists whether ANSI mode is on or off, as shown below,

>>> pser = pd.Series([1.1, 2.2, 3.3], dtype=np.float32)
>>> psser = ps.from_pandas(pser)
>>> spark.conf.set("spark.sql.ansi.enabled", False)
>>> psser / 1
0    1.1                                                                        
1    2.2
2    3.3
dtype: float64
>>> spark.conf.set("spark.sql.ansi.enabled", True)
>>> psser / 1
0    1.1
1    2.2
2    3.3
dtype: float64

Does this PR introduce any user-facing change?

Yes. truediv/floordiv under ANSI works as expected, with ANSI on/off, as shown below.

>>> import pandas as pd
>>> import numpy as np
>>> 
>>> ps.set_option("compute.fail_on_ansi_mode", False)
>>> ps.set_option("compute.ansi_mode_support", True)
>>> 
>>> pser = pd.Series([1.1, 2.2, 3.3], dtype=np.float32)
>>> psser = ps.from_pandas(pser)
>>> psser / 1
0    1.1                                                                        
1    2.2
2    3.3
dtype: float32
>>> psser // 1
0    1.0
1    2.0
2    3.0
dtype: float32

How was this patch tested?

Unit tests.

Commands below all passed

 1103  SPARK_ANSI_SQL_MODE=true  ./python/run-tests --python-executables=python3.11 --testnames "pyspark.pandas.tests.computation.test_binary_ops FrameBinaryOpsTests.test_binary_operator_truediv"
 1104  SPARK_ANSI_SQL_MODE=false  ./python/run-tests --python-executables=python3.11 --testnames "pyspark.pandas.tests.computation.test_binary_ops FrameBinaryOpsTests.test_binary_operator_truediv"
 1106  SPARK_ANSI_SQL_MODE=true  ./python/run-tests --python-executables=python3.11 --testnames "pyspark.pandas.tests.computation.test_binary_ops FrameBinaryOpsTests.test_binary_operator_floordiv"
 1108  SPARK_ANSI_SQL_MODE=false  ./python/run-tests --python-executables=python3.11 --testnames "pyspark.pandas.tests.computation.test_binary_ops FrameBinaryOpsTests.test_binary_operator_floordiv"
 1126  git status
 1127  SPARK_ANSI_SQL_MODE=true  ./python/run-tests --python-executables=python3.11 --testnames "pyspark.pandas.tests.computation.test_binary_ops FrameBinaryOpsTests.test_divide_by_zero_behavior"
 1128  SPARK_ANSI_SQL_MODE=false  ./python/run-tests --python-executables=python3.11 --testnames "pyspark.pandas.tests.computation.test_binary_ops FrameBinaryOpsTests.test_divide_by_zero_behavior"

Was this patch authored or co-authored using generative AI tooling?

No.