Validate FeatureVectorizer batch sizes (#31671)
This pull request improves the robustness of the `FeatureVectorizer`
operator by enforcing that all input tensors have the same batch size
and adds a corresponding unit test to verify this behavior.
**FeatureVectorizer input validation:**
* Added a check in `feature_vectorizer.cc` to ensure all input tensors
to `FeatureVectorizer` have the same batch size, raising an error if a
mismatch is detected.
**Unit test improvements:**
* Added a new test `RejectMismatchedBatchSizes` in
`feature_vectorizer_test.cc` to verify that the operator correctly
rejects inputs with mismatched batch sizes.
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
