Enable ruff pandas ruleset

Unfortunately with bugs in ruff (e.g. 2143) that requires noqa
statements where they really should not be - ruff mistakes polars
objects as pandas objects.

Author: berland

pyproject.toml

@@ -216,6 +216,7 @@ select = [
     "ISC",
     "LOG",   # logging
     "NPY",   # numpy specific rules
+    "PD",
     "PIE",   # flake8-pie
     "PL",    # pylint
     "PLE",

src/ert/dark_storage/endpoints/responses.py

@@ -177,7 +177,7 @@ def data_for_response(
             assert "report_step" in filter_on
             report_step = int(filter_on["report_step"])
             vals = data.filter(pl.col("report_step").eq(report_step))
-            pivoted = vals.drop("response_key", "report_step").pivot(
+            pivoted = vals.drop("response_key", "report_step").pivot(  # noqa: PD010
                 on="index", values="values"
             )
             data = pivoted.to_pandas().set_index("realization")

src/ert/plugins/hook_implementations/workflows/csv_export.py

@@ -111,7 +111,7 @@ def run(
                 summary_data = pl.DataFrame({})
 
             if not summary_data.is_empty():
-                pivoted_summary = summary_data.pivot(
+                pivoted_summary = summary_data.pivot(  # noqa: PD010
                     index=["realization", "time"], on="response_key", values="values"
                 ).to_pandas()
 

src/ert/storage/local_ensemble.py

@@ -1054,7 +1054,7 @@ def get_observations_and_responses(
                                 pl.col(col).is_in(observed_values.implode())
                             )
 
-                    pivoted = responses.collect(engine="streaming").pivot(
+                    pivoted = responses.collect(engine="streaming").pivot(  # noqa: PD010
                         on="realization",
                         index=["response_key", *response_cls.primary_key],
                         values="values",
@@ -1230,7 +1230,7 @@ def all_parameters_and_gen_data(self) -> pl.DataFrame | None:
             how="horizontal",
         )
 
-        responses_wide = responses["realization", "response_key", "values"].pivot(
+        responses_wide = responses["realization", "response_key", "values"].pivot(  # noqa: PD010
             on="response_key", values="values"
         )
 

tests/ert/performance_tests/test_dark_storage_performance.py

@@ -103,7 +103,7 @@ def get_record_observations(storage, ensemble_id, keyword: str, poly_ran):
             assert len(obs) == num_summary_obs
             assert np.isclose(obs[0].errors[0], 0.1)
             assert obs[0].x_axis[0].startswith("2010-01-02T00:00:00")
-            assert np.isclose(obs[0].values[0], 2.6357)
+            assert np.isclose(obs[0].values[0], 2.6357)  # noqa: PD011
             assert len(obs[0].errors) == 1
             assert len(obs[0].x_axis) == 1
             assert len(obs[0].values) == 1