Tidy spandrel model probe logic, and document the reasons behind the current implementation.

invokeai/backend/model_manager/probe.py

@@ -4,6 +4,7 @@ from pathlib import Path
 from typing import Any, Dict, Literal, Optional, Union
 
 import safetensors.torch
+import spandrel
 import torch
 from picklescan.scanner import scan_file_path
 
@@ -242,15 +243,19 @@ class ModelProbe(object):
             return ModelType.TextualInversion
 
         # Check if the model can be loaded as a SpandrelImageToImageModel.
+        # This check is intentionally performed last, as it can be expensive (it requires loading the model from disk).
         try:
-            # TODO(ryand): Figure out why load_from_state_dict() doesn't work as expected.
-            # _ = SpandrelImageToImageModel.load_from_state_dict(ckpt)
+            # It would be nice to avoid having to load the Spandrel model from disk here. A couple of options were
+            # explored to avoid this:
+            # 1. Call `SpandrelImageToImageModel.load_from_state_dict(ckpt)`, where `ckpt` is a state_dict on the meta
+            #    device. Unfortunately, some Spandrel models perform operations during initialization that are not
+            #    supported on meta tensors.
+            # 2. Spandrel has internal logic to determine a model's type from its state_dict before loading the model.
+            #    This logic is not exposed in spandrel's public API. We could copy the logic here, but then we have to
+            #    maintain it, and the risk of false positive detections is higher.
             _ = SpandrelImageToImageModel.load_from_file(model_path)
             return ModelType.SpandrelImageToImage
-        except Exception as e:
-            # TODO(ryand): Catch a more specific exception type here if we can.
-            # TODO(ryand): Delete this print statement.
-            print(e)
+        except spandrel.UnsupportedModelError:
             pass
 
         raise InvalidModelConfigException(f"Unable to determine model type for {model_path}")