fix: Removed units from TimeSeriesBase model and changed Pint

behaviour.

List of changes
- Changed data handling when we create an instance of
  `SingleTimeSeries`.
- Updated test to check that data inside a `SingleTimeSeries` is consistent since it could return
  two different instances when serialize and deserialize.
- Updated arrow storage to convert to `pa.Array` when serializing only
  and better type hint

src/infrasys/arrow_storage.py

@@ -123,13 +123,15 @@ def _get_single_time_series(
             normalization=metadata.normalization,
         )
 
-    def _convert_to_record_batch(self, array: SingleTimeSeries, variable_name: str):
+    def _convert_to_record_batch(
+        self, time_series: SingleTimeSeries, variable_name: str
+    ) -> pa.RecordBatch:
         """Create record batch to save array to disk."""
-        pa_array = array.data.magnitude if isinstance(array.data, BaseQuantity) else array.data
-        if not isinstance(array.data, pa.Array) and isinstance(
-            array.data, BaseQuantity | pint.Quantity
+        pa_array = time_series.data
+        if not isinstance(pa_array, pa.Array) and isinstance(
+            pa_array, BaseQuantity | pint.Quantity
         ):
-            pa_array = pa.array(array.data.magnitude)
+            pa_array = pa.array(pa_array.magnitude)
         assert isinstance(pa_array, pa.Array)
         schema = pa.schema([pa.field(variable_name, pa_array.type)])
         return pa.record_batch([pa_array], schema=schema)

src/infrasys/time_series_models.py

@@ -45,7 +45,6 @@ class TimeSeriesStorageType(str, Enum):
 class TimeSeriesData(InfraSysBaseModelWithIdentifers, abc.ABC):
     """Base class for all time series models"""
 
-    units: Optional[str] = None
     variable_name: str
     normalization: NormalizationModel = None
 
@@ -74,16 +73,15 @@ def length(self) -> int:
 
     @field_validator("data", mode="before")
     @classmethod
-    def check_data(cls, data) -> pa.Array | BaseQuantity:  # Standarize what object we receive.
+    def check_data(
+        cls, data
+    ) -> pa.Array | pa.ChunkedArray | pint.Quantity:  # Standarize what object we receive.
         """Check time series data."""
         if len(data) < 2:
             msg = f"SingleTimeSeries length must be at least 2: {len(data)}"
             raise ValueError(msg)
 
-        if isinstance(data, BaseQuantity | pint.Quantity):
-            if not isinstance(data.magnitude, pa.Array):
-                cls = type(data)
-                return cls(data.magnitude, data.units)
+        if isinstance(data, pint.Quantity | BaseQuantity):
             return data
 
         if not isinstance(data, pa.Array):

tests/test_single_time_series.py

@@ -82,8 +82,8 @@ def test_with_quantity():
     assert ts.length == length
     assert ts.resolution == resolution
     assert ts.initial_time == initial_time
-    assert isinstance(ts.data.magnitude, pa.Array)
-    assert ts.data[-1].as_py() == length - 1
+    assert isinstance(ts.data, ActivePower)
+    assert ts.data[-1].magnitude == length - 1
 
 
 def test_normalization():

tests/test_system.py

@@ -249,10 +249,11 @@ def test_time_series():
 
 
 @pytest.mark.parametrize(
-    "params", list(itertools.product([True, False], [True, False], [True, False]))
+    "in_memory,use_quantity,sql_json",
+    list(itertools.product([True, False], [True, False], [True, False])),
 )
-def test_time_series_retrieval(params):
-    in_memory, use_quantity, sql_json = params
+def test_time_series_retrieval(in_memory, use_quantity, sql_json):
+    # in_memory, use_quantity, sql_json = params
     try:
         if not sql_json:
             os.environ["__INFRASYS_NON_JSON_SQLITE__"] = "1"
@@ -290,29 +291,33 @@ def test_time_series_retrieval(params):
         for metadata in system.list_time_series_metadata(gen, scenario="high"):
             assert metadata.user_attributes["scenario"] == "high"
 
-        assert (
-            system.get_time_series(
-                gen, variable_name=variable_name, scenario="high", model_year="2030"
+        assert all(
+            np.equal(
+                system.get_time_series(
+                    gen, variable_name, scenario="high", model_year="2030"
+                ).data,
+                ts1.data,
             )
-            == ts1
         )
-        assert (
-            system.get_time_series(
-                gen, variable_name=variable_name, scenario="high", model_year="2035"
+        assert all(
+            np.equal(
+                system.get_time_series(
+                    gen, variable_name, scenario="high", model_year="2035"
+                ).data,
+                ts2.data,
             )
-            == ts2
         )
-        assert (
-            system.get_time_series(
-                gen, variable_name=variable_name, scenario="low", model_year="2030"
+        assert all(
+            np.equal(
+                system.get_time_series(gen, variable_name, scenario="low", model_year="2030").data,
+                ts3.data,
             )
-            == ts3
         )
-        assert (
-            system.get_time_series(
-                gen, variable_name=variable_name, scenario="low", model_year="2035"
+        assert all(
+            np.equal(
+                system.get_time_series(gen, variable_name, scenario="low", model_year="2035").data,
+                ts4.data,
             )
-            == ts4
         )
 
         with pytest.raises(ISAlreadyAttached):
@@ -324,12 +329,6 @@ def test_time_series_retrieval(params):
             gen, variable_name=variable_name, scenario="high", model_year="2030"
         )
         assert not system.has_time_series(gen, variable_name=variable_name, model_year="2036")
-        assert (
-            system.get_time_series(
-                gen, variable_name=variable_name, scenario="high", model_year="2030"
-            )
-            == ts1
-        )
         with pytest.raises(ISOperationNotAllowed):
             system.get_time_series(gen, variable_name=variable_name, scenario="high")
         with pytest.raises(ISNotStored):