feat(sayt): Add SAYTBuilder that constructs runtime artefacts for later use

[Tom-Owen-ONS]

📌 Pull Request Template

Please complete all sections

✨ Summary

This PR adds a SAYTBuilder class, which takes a set of configured RetrieverSpecs, as well as a corpus and constructs an artefact folder that contains the built vector db parquet files, as well as the corpus and a manifest to locate all of the necessary files to run a SAYTSuggester. It allows for building of the artefacts at image build stage, then loading from the artefact directory from a SAYTSuggester.from_artefact() method.

📜 Changes Introduced

• Add SAYTBuilder to build an artefact directory to construct a SAYTSuggester quickly from a pre-build vector db.

• Adds an example notebook to test this functionality

• Expands SAYTSuggester.get_config() to return a SaytConfiguration Pydantic model, which contains information about global suggester settings, the corpus, the retrievers and the artefact provenance (file locations etc).

• Feature implementation (feat:) / bug fix (fix:) / refactoring (chore:) / documentation (docs:) / testing (test:)

• Updates to tests and/or documentation

• Terraform changes (if applicable)

✅ 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

A notebook has been provided in demos/sayt/sayt_artifact_example.py to show the basic usage of SAYTBuilder.

[Tom-Owen-ONS]

For consideration: remove the sayt_* prefixes from the modules in src/industrial_classification_utils/sayt, as they're already in a sub-package called 'sayt' so it should implied?

[Tom-Owen-ONS]

This does not directly implement the use of min_chars per retriever as it was already getting quite big and it would be nice to add this as a feature with separate example notebooks etc. Happy to be challenged on this and add it here though

[ivyONS]

Works as expected, provides good speedup on load.

Next steps:
Async: Consider building/loading/querying the different retrievers in parallel.
Can we add caching? (as the user types more characters the initial part of the string will be searched repeatedly)

[ivyONS]

LGTM