Implement BiRefNet and HAT Sharper models with background removal capabilities

- Added validation documentation for AI model tiers.
- Introduced `background.py` for background removal logic, including inference functions for alpha and foreground extraction.
- Created `onnx_birefnet.py` to handle BiRefNet model sessions and alpha inference.
- Developed tests for background models, ensuring proper dispatch and session reuse.
- Added tests for BiRefNet ONNX model to validate alpha output and error handling for invalid values.
- Implemented model export scripts for BiRefNet and HAT Sharper, ensuring proper weight compression and licensing.
- Created `deform_sampling.py` for deformable convolution operations used in BiRefNet.
- Added utility functions for compressing ONNX weights without altering graph arithmetic.
- Introduced a reference plane node for depth field adjustments in geometry processing.

tests/anyimage/operators/test_texture_node_actions.py

@@ -113,7 +113,7 @@ def test_texture_action_undo_redo_restores_image_and_binding(texture, tmp_path,
         node = bpy.data.materials["Texture material"].node_tree.nodes["Source texture"]
         operator._image_target = ImageEditTarget.capture(texture_context(node))
         if operation == upscale.UpscaleImage:
-            operator.model = "PISA_SR"
+            operator.model = "HAT_GAN_X4_SHARPER"
         operator.response(context, SimpleNamespace(directory=tmp_path))
         return {"FINISHED"}
 
@@ -175,7 +175,7 @@ def test_texture_job_request_and_completion(texture, tmp_path, operation, module
             stack.enter_context(patch.object(runtime, "update_ui"))
             stack.enter_context(patch.object(runtime, "redraw_ui"))
             if operation == upscale.UpscaleImage:
-                stack.enter_context(patch.object(upscale, "configured_upscale_model", return_value="SAFMN_REAL_X4"))
+                stack.enter_context(patch.object(upscale, "configured_upscale_model", return_value="HAT_GAN_X4_SHARPER"))
                 stack.enter_context(patch.object(upscale, "configured_max_ai_input_size", return_value=2048))
             assert original_execute(operator, context) == {"RUNNING_MODAL"}
             job = runtime.active_job
@@ -185,7 +185,7 @@ def test_texture_job_request_and_completion(texture, tmp_path, operation, module
             assert kind == operation.job_type
             assert parameters["start_frame"] == 1
             assert parameters["device"] == "cpu"
-            assert parameters["model"] == ("SAFMN_REAL_X4" if operation == upscale.UpscaleImage else "BEN2_BASE")
+            assert parameters["model"] == ("HAT_GAN_X4_SHARPER" if operation == upscale.UpscaleImage else "BIREFNET_LITE")
             input_path = operator.input_path
             assert operator.delete_input
             # Selection changes must not redirect completion.