Fix Pydantic validator mutating caller's dict (#661)

The collect_additional_datasets model validator was modifying the input
dict in-place, adding an empty 'other_datasets' key. When users reused
their data_request dict (e.g., adding 'infer' after setting 'train'),
the empty 'other_datasets' would cause DataProvider to fail with
"No datasets were requested in data_request".

Co-authored-by: Claude Opus 4.5 <[email protected]>

Author: gitosaurus

src/hyrax/config_schemas/data_request.py

@@ -113,6 +113,8 @@ def normalize_dataset_configs(cls, value: Any) -> Any:
     def collect_additional_datasets(cls, values: dict[str, Any]) -> dict[str, Any]:
         """Capture arbitrary dataset keys beyond train/validate/infer."""
 
+        # Copy to avoid mutating the caller's dict
+        values = dict(values)
         known = {"train", "validate", "infer"}
         extra = {k: v for k, v in values.items() if k not in known}
         for key in extra: