Added TA-SQL schema linking module

[fi5421]

Description

This PR corresponds to the following Task

Added TASL (schema linking module from TA-SQL) into schema_linking_utils. Currently, it only utilizes the question_id from dev.json to find all relevant details such as evidence, db_id. It gathers schema information mostly from the dev_tables.json file. I have tried to keep the tasl.py module as close to the original as possible right now, this includes some comments from the original code. Changes in the module were required to use our dataset structure.

Currently this does not yield good results but if we want to use it in the main pipeline we could pass everything as parameters in the function rather that it fetching from the dataset.

This PR also adds clients for gemini 2.5 pro preview and o4 (from the main branch)

[Copilot]

Pull Request Overview

This PR introduces a new TA‐SQL schema linking module (TASL) and integrates it into the existing schema linking utilities, while also adding new client support for Gemini models.

• Added the TASL module under server/utilities/schema_linking/tasl.py for schema linking based on dev.json and dev_tables.json.
• Updated schema_linking_utils.py to include a new tasl function and integrated TASL usage.
• Extended LLM_enums.py with additional model types to support new Gemini clients.

Reviewed Changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.

File	Description
server/utilities/schema_linking/tasl.py	Added the TASL module with schema reconstruction and dummy SQL generation.
server/utilities/schema_linking/schema_linking_utils.py	Introduced a new tasl function that instantiates TASL.
server/utilities/constants/LLM_enums.py	Updated enum values with new Gemini model types.

[fi5421]

Updated the TASL schema linking dummy prompt to be used from the prompt factory.
The prompt uses column_meaning.json for descriptions in the schema.
It also needed a primary_keys dicts so created another function in utility_functions.py to get a primary key dict.
Dummy SQL requires target_question, evidence and database_name in the get_prompt_class function only.