feat: Implement weight cutout boundary smoothing for image depth cutout

- Added a new proposal for boundary smoothing in the weight cutout process, differentiating between Split edges and Outline smoothing requirements.
- Introduced a new capability `cutout-boundary-smoothing` to control smoothing strength and default values.
- Updated specifications to define the behavior of boundary-specific smoothing strength and default settings.
- Created tasks to verify the implementation of boundary behavior and asset updates.
- Developed comprehensive tests for depth symmetry and normal scaling to ensure expected behavior and performance.
- Introduced new attributes for normal mapping to facilitate persistent face attributes sharing between depth symmetry and image materials.
- Built a new node group for O Image Depth Symmetry, integrating the new smoothing capabilities and ensuring compatibility with existing workflows.

tests/tools/nodes/test_cutouts.py

@@ -13,7 +13,7 @@ from anyimage.common.object import modifier_input_identifier, set_modifier_input
 
 
 @pytest.mark.parametrize("source", ("definition", "asset"))
-@pytest.mark.parametrize("shape", ("SOLID", "DEPTH_SOLID", "DEPTH_BALLOON"))
+@pytest.mark.parametrize("shape", ("SOLID", "DEPTH_SOLID", "DEPTH_SYMMETRY"))
 def test_depth_axis_is_visible_in_options(shape, source):
     bpy.ops.wm.read_factory_settings(use_empty=True)
     group = CUTOUT_BUILDERS[shape]() if source == "definition" else load_cutout_node_group(shape)
@@ -55,7 +55,7 @@ def test_cutout_thickness_modes():
             assert np.isfinite(z).all() and len(z)
             if thickness == 0:
                 assert len(result.polygons) == 4 and np.max(abs(z)) < 1e-7
-                assert all(abs(item.value - 1) < 1e-7 for item in result.attributes["o_normal_scale"].data)
+                assert all(abs(item.value) < 1e-7 for item in result.attributes["o_normal_reduction"].data)
             else:
                 back, front = (-0.6, 0.6) if mode == 0 else (-thickness, 0.0)
                 assert np.isclose(z.min(), back) and np.isclose(z.max(), front)