Fix issue where moosez breaks if a 'moosez' prefixed folder exists as…

… the subject was already processed and segmentations exist.

Before this change, running `moosez` would fail if there was already a folder with a name starting with 'moosez'. This commit makes `moosez` ignore such folders, allowing multiple models to run sequentially without manual folder cleanup.

This patch allows commands like the following to be run one after another without any issues:
- `moosez -d path_to_dir -m your_model_1`
- `moosez -d path_to_dir -m your_model_2`

moosez/image_conversion.py

@@ -117,11 +117,10 @@ def standardize_to_nifti(parent_dir: str) -> None:
             if os.path.isdir(subject_path):
                 images = os.listdir(subject_path)
                 for image in images:
-                    if os.path.isdir(os.path.join(subject_path, image)):
-                        image_path = os.path.join(subject_path, image)
-                        non_nifti_to_nifti(image_path)
-                    elif os.path.isfile(os.path.join(subject_path, image)):
-                        image_path = os.path.join(subject_path, image)
+                    image_path = os.path.join(subject_path, image)
+                    path_is_valid = os.path.isdir(image_path) | os.path.isfile(image_path)
+                    path_is_valid = path_is_valid & ("moosez" not in os.path.basename(image_path))
+                    if path_is_valid:
                         non_nifti_to_nifti(image_path)
             else:
                 continue