feat(sql_database): support content_and_artifact in SQLDatabase.run

[diego-coder]

Description

This pull request addresses issue #30179 by adding support for the response_format="content_and_artifact" option in the SQLDatabase.run() method.

Previously, tools like QuerySQLDatabaseTool raised a ValueError when this response format was requested, because the underlying run() method always returned a string. With this change, users can now retrieve both the LLM-friendly string output and a structured Python object for programmatic use unblocking important downstream use cases.

Changes included in this PR:

1. langchain_community/utilities/sql_database.py:

   • The run() method signature now accepts **kwargs (for backward compatibility and future extensibility).
   • Added logic to check for a response_format keyword; when set to "content_and_artifact", the method returns a tuple (string_result, processed_result).
   • The artifact returned is the post-processed result (with truncated strings), matching the structure of what is returned to the LLM in string form.
   • Updated the method docstring to Google-style.
   • Backward compatibility: Existing code using run() without response_format will see no change in behavior.

2. langchain_community/tests/unit_tests/utilities/test_sql_database.py:

   • Created a new unit test file for the core SQLDatabase utility (previously missing).
   • Added test_sql_database_run_content_and_artifact to validate both the new tuple-based return and the default string format. This ensures correct behavior and prevents regressions.

Fixes #30179