BUG: AI tagging hangs indefinitely when an image fails to load, blocking subsequent tagging jobs

### Is there an existing issue for this?

- [x] I have searched the existing issues

### What happened?

## Description
While AI tagging, if the backend encounters an image that it cannot load (cv2.imread() returns None), the entire AI-tagging job stops instead of skipping the image. The folder remains in the "processing" state indefinitely, and any following queued folder also remain pending until the ai tagging process of the folder that has problematic image, is turned off. Still we need to restart the subsequent folder's ai tagging.

### What backend logs:-

Failed to load image: -path-
Error processing untagged images: object of type 'NoneType' has no len()

problematic image:-

properties: -
format: JPEG
resolution: 616 x 920
size: 79kb

### Current behavior

https://github.com/user-attachments/assets/95be7292-6dce-4393-8641-b4f79dc12696

### Expected behavior
- If an image cannot be loaded, log the error.
- Skip the problematic image.
- Continue processing the remaining images.
- Continue processing later queued folders.

### Record

- [x] I agree to follow this project's Code of Conduct

Start at the backend AI-tagging path that calls cv2.imread() and produces the reported load and NoneType errors. Trace how one failed image affects the current folder's processing state and the queued folders, then use the existing AI-tagging tests or reproduce with the reported image. Done means the failure is logged, that image is skipped, remaining images are processed, and later folders continue.

由索引模型根据 Issue 内容生成。