Fix transformed labels with outline

tests/pl/test_render_labels.py

@@ -9,6 +9,7 @@
 from spatial_image import to_spatial_image
 from spatialdata import SpatialData, deepcopy, get_element_instances
 from spatialdata.models import TableModel
+from spatialdata.transformations import Affine, Scale, Sequence, Translation, set_transformation
 
 import spatialdata_plot  # noqa: F401
 from tests.conftest import DPI, PlotTester, PlotTesterMeta
@@ -194,3 +195,52 @@ def test_plot_subset_categorical_label_maintains_order_when_palette_overwrite(se
         sdata_blobs.pl.render_labels(
             "blobs_labels", color="which_max", groups=["channel_0_sum"], palette="red"
         ).pl.show(ax=axs[1])
+
+    def test_plot_can_render_transformed_labels_with_outline(self, sdata_blobs: SpatialData):
+        # Blobs have by default an identity transform. Replace it by a rotated, scaled transform:
+        angle = np.deg2rad(15)
+        axes = ("y", "x")
+        rotation = Affine(
+            matrix=np.array(
+                [
+                    [np.cos(angle), -np.sin(angle), 0],
+                    [np.sin(angle), np.cos(angle), 0],
+                    [0, 0, 1],
+                ]
+            ),
+            input_axes=axes,
+            output_axes=axes,
+        )
+        translation = Translation([10, 10], axes=axes)
+        scale = Scale([0.65, 0.65], axes=axes)
+        transform = Sequence([translation, rotation, scale])
+        set_transformation(sdata_blobs["blobs_labels"], transform)
+
+        # Render without fill, but with outline.
+        sdata_blobs.pl.render_labels(
+            "blobs_labels",
+            fill_alpha=0.0,
+            outline=True,
+            outline_alpha=1.0,
+        ).pl.show()
+
+    def _make_tablemodel_with_categorical_labels(self, sdata_blobs, label):
+        n_obs = max(_get_unique_label_values_as_index(sdata_blobs[label]))
+        adata = AnnData(
+            RNG.normal(size=(n_obs, 10)),
+            obs=pd.DataFrame(RNG.normal(size=(n_obs, 3)), columns=["a", "b", "c"]),
+        )
+        adata.obs["instance_id"] = np.arange(adata.n_obs)
+        adata.obs["category"] = RNG.choice(["a", "b", "c"], size=adata.n_obs)
+        adata.obs["category"][:3] = ["a", "b", "c"]
+        adata.obs["instance_id"] = list(range(adata.n_obs))
+        adata.obs["region"] = label
+        table = TableModel.parse(
+            adata=adata,
+            region_key="region",
+            instance_key="instance_id",
+            region=label,
+        )
+        sdata_blobs["other_table"] = table
+        sdata_blobs["other_table"].obs["category"] = sdata_blobs["other_table"].obs["category"].astype("category")
+        sdata_blobs.pl.render_labels(label, color="category").pl.show()