Fix transient test failure in test_BalanceDFWeights_trim (#225)

Summary:
- [x] Analyze the failing test `TestBalanceDFWeights.test_BalanceDFWeights_trim`
- [x] Understand the root cause: test uses random data and expects deterministic threshold
- [x] Identify that the 88.98th percentile can occasionally exceed 0.9 (~0.037% probability)
- [x] Fix the test by adding `np.random.seed(112358)` for reproducibility (consistent with other tests)
- [x] Run the test to verify the fix (passes consistently)
- [x] Remove debug print statements from test (code review feedback)
- [x] Use consistent seed value across the test file (code review feedback)
- [x] Run final code review and security checks (no issues found)
- [x] Run all tests in test_balancedf.py (83 passed)
- [x] Remove auto-generated version.py file and add to .gitignore (PR feedback)

Fixed the transient test failure by adding a fixed random seed (112358) for reproducibility and removing debug print statements. Removed auto-generated version.py file that was accidentally committed.

<details>

<summary>Original prompt</summary>

----

*This section details on the original issue you should resolve*

<issue_title>[BUG] Fix transient issue in TestBalanceDFWeights.test_BalanceDFWeights_trim</issue_title>
<issue_description>log

```
============================ test session starts ==============================
platform linux -- Python 3.12.12, pytest-9.0.2, pluggy-1.6.0
rootdir: /home/runner/work/balance/balance
configfile: pyproject.toml
plugins: anyio-4.12.0
collected 501 items

tests/test_adjust_null.py .                                              [  0%]
tests/test_adjustment.py .......................................         [  7%]
tests/test_balancedf.py ..................F............................. [ 17%]
...................................                                      [ 24%]
tests/test_cbps.py ....................                                  [ 28%]
tests/test_cli.py .....................                                  [ 32%]
tests/test_datasets.py .............                                     [ 35%]
tests/test_general_stats.py .........                                    [ 37%]
tests/test_ipw.py ...........................                            [ 42%]
tests/test_logging.py ...........                                        [ 44%]
tests/test_poststratify.py ......                                        [ 45%]
tests/test_rake.py .......................                               [ 50%]
tests/test_sample.py ................................................... [ 60%]
........................                                                 [ 65%]
tests/test_sample_diagnostics_helper.py ..............                   [ 68%]
tests/test_stats_and_plots.py .......................................... [ 76%]
...                                                                      [ 77%]
tests/test_testutil.py ..................................                [ 84%]
tests/test_util.py ..................................................... [ 94%]
...............                                                          [ 97%]
tests/test_weighted_comparisons_plots.py ............                    [100%]

=================================== FAILURES ===================================
_______________ TestBalanceDFWeights.test_BalanceDFWeights_trim ________________

self = <test_balancedf.TestBalanceDFWeights testMethod=test_BalanceDFWeights_trim>

    def test_BalanceDFWeights_trim(self) -> None:
        s = Sample.from_frame(
            pd.DataFrame({"w": np.random.uniform(0, 1, 10000), "id": range(0, 10000)}),
            id_column="id",
            weight_column="w",
        )
        s.weights().trim(percentile=(0, 0.11), keep_sum_of_weights=False)
        print(s.weights().df)
        print(max(s.weights().df.iloc[:, 0]))
>       self.assertTrue(max(s.weights().df.iloc[:, 0]) < 0.9)
E       AssertionError: False is not true

tests/test_balancedf.py:483: AssertionError
----------------------------- Captured stdout call -----------------------------
             w
0     0.615525
1     0.902178
2     0.707789
3     0.527844
4     0.623877
...        ...
9995  0.902178
9996  0.655653
9997  0.430850
9998  0.110247
9999  0.282947

[10000 rows x 1 columns]
0.9021778137132038
----------------------------- Captured stderr call -----------------------------
WARNING (2025-12-22 18:31:49,545) [sample_class/from_frame (line 432)]: Casting id column to string
WARNING (2025-12-22 18:31:49,554) [util/_warn_of_df_dtypes_change (line 2239)]: The dtypes of sample._df were changed from the original dtypes of the input df, here are the differences -
WARNING (2025-12-22 18:31:49,555) [util/_warn_of_df_dtypes_change (line 2248)]: The (old) dtypes that changed for df (before the change):
WARNING (2025-12-22 18:31:49,555) [util/_warn_of_df_dtypes_change (line 2251)]:
id    int64
dtype: object
WARNING (2025-12-22 18:31:49,555) [util/_warn_of_df_dtypes_change (line 2252)]: The (new) dtypes saved in df (after the change):
WARNING (2025-12-22 18:31:49,556) [util/_warn_of_df_dtypes_change (line 2253)]:
id    object
dtype: object
------------------------------ Captured log call -------------------------------
WARNING  balance:sample_class.py:432 Casting id column to string
WARNING  balance:util.py:2239 The dtypes of sample._df were changed from the original dtypes of the input df, here are the differences -
WARNING  balance:util.py:2248 The (old) dtypes that changed for df (before the change):
WARNING  balance:util.py:2251
id    int64
dtype: object
WARNING  balance:util.py:2252 The (new) dtypes saved in df (after the change):
WARNING  balance:util.py:2253
id    object
dtype: object
=========================== short test summary info ============================
FAILED tests/test_balancedf.py::TestBalanceDFWeights::test_BalanceDFWeights_trim - AssertionError: False is not true
================== 1 failed, 500 passed in 300.92s (0:05:00) ===================
Error: Process completed with exit code 1.

```</issue_description>

<ag...

</details>

- Fixes #224

 ---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our [2 minute survey](https://gh.io/copilot-coding-agent-survey).

Pull Request resolved: #225

Reviewed By: omriharosh

Differential Revision: D89710130

Pulled By: talgalili

fbshipit-source-id: 8c2a8bd4ddcecdc9510032e1e6d9373ea0e4dbf9

Author: Copilot

.gitignore

@@ -1,6 +1,9 @@
 # Byte-compiled / optimized / DLL files
 __pycache__/
 
+# Auto-generated version file
+version.py
+
 # OSX
 *.DS_Store
 

tests/test_balancedf.py

@@ -472,14 +472,13 @@ def test_BalanceDFWeights_design_effect(self) -> None:
         self.assertTrue(s.weights().design_effect(), 7 / 3)
 
     def test_BalanceDFWeights_trim(self) -> None:
+        np.random.seed(112358)  # Fix seed for reproducibility
         s = Sample.from_frame(
             pd.DataFrame({"w": np.random.uniform(0, 1, 10000), "id": range(0, 10000)}),
             id_column="id",
             weight_column="w",
         )
         s.weights().trim(percentile=(0, 0.11), keep_sum_of_weights=False)
-        print(s.weights().df)
-        print(max(s.weights().df.iloc[:, 0]))
         self.assertTrue(max(s.weights().df.iloc[:, 0]) < 0.9)
 
     def test_BalanceDFWeights_summary(self) -> None: