[QoL fix] [Image processing] Add warning on assumption of channel dim and...

[QoL fix] [Image processing] Add warning on assumption of channel dim and avoid infering when inputs are PIL.Image (#31364)

* Add warning on assumption of channel dim
Use PIL info whenever possible to decide channel axis

* Fix ruff format

* Remove type checking
Improve warning message

* Update src/transformers/models/siglip/image_processing_siglip.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/image_utils.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/image_utils.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

---------
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

src/transformers/image_utils.py

@@ -234,7 +234,12 @@ def infer_channel_dimension_format(
     else:
         raise ValueError(f"Unsupported number of image dimensions: {image.ndim}")
 
-    if image.shape[first_dim] in num_channels:
+    if image.shape[first_dim] in num_channels and image.shape[last_dim] in num_channels:
+        logger.warning(
+            f"The channel dimension is ambiguous. Got image shape {image.shape}. Assuming channels are the first dimension."
+        )
+        return ChannelDimension.FIRST
+    elif image.shape[first_dim] in num_channels:
         return ChannelDimension.FIRST
     elif image.shape[last_dim] in num_channels:
         return ChannelDimension.LAST