SA-663 Load embedding from flat file

[ivyONS]

✨ Summary

Refactors EmbeddingHandler to load and build vector stores from arbitrary flat CSV files instead of only from the SIC hierarchy XLS sources. SIC-specific loading is extracted into a standalone utility (sic_specific_embed.py). Return types for search_index and search_index_multi are replaced with typed Pydantic response models, that is imported into ***-vector-store-api. GCS support is extended to single-file downloads (to handle CSV indes source files).

📜 Changes Introduced

• feat: EmbeddingHandler.__init__ accepts index_source_file (CSV path or GCS URI); when provided, builds a new vector store; otherwise loads an existing one — replaces the previous _load_or_build_vector_store logic
• feat: _load_existing_vector_store now raises FileNotFoundError instead of returning None, making the error explicit
• feat: _build_vector_store — loads data from flat file (native classifai format, no metadata)
• feat: New sic_specific_embed.load_embedding_handler_from_sic_index_files preserves the old SIC-hierarchy-based build path as an explicit utility function
• feat: gcs_file_access.download_one_file_from_gcs — downloads a single GCS file to a temp directory for use during vector store build
• feat: SearchIndexItem & SearchIndexResponse Pydantic models replace list[dict] return types on search_index and search_index_multi
• feat: EmbeddingHandler.get_embed_config returns a typed EmbeddingConfig snapshot
• chore: EmbeddingConfig migrated from TypedDict to Pydantic BaseModel; adds index_source_file and index_size fields
• chore: Example data files converted from .txt to .csv (toy_index, sic_2d_condensed, sic_4d_condensed)
• chore: CustomVertexAIEmbeddings removed
• test: test_embedding.py fully updated - adds unit tests
• test: test_gcs_file_access.py adds tests for download_one_file_from_gcs (success + missing file)

✅ Checklist

Please confirm you've completed these checks before requesting a review.

• Code is formatted using Black
• Imports are sorted using isort
• Code passes linting with Ruff, Pylint, and Mypy
• Security checks pass using Bandit
• API and Unit tests are written and pass using pytest
• Terraform files (if applicable) follow best practices and have been validated (terraform fmt & terraform validate)
• DocStrings follow Google-style and are added as per Pylint recommendations
• Documentation has been updated if needed

🔍 How to Test

• check all test pass make all-tests
• check demo notebook (demos/embed/sic_embedding_example.py)
• make sure it works in sic-classification-vector-store