fix: merge datascience v3 and v2

[peteryang1]

Description

Motivation and Context

How Has This Been Tested?

• If you are adding a new feature, test on your own test scripts.

Screenshots of Test Results (if appropriate):

1. Your own tests:

Types of changes

• Fix bugs
• Add new feature
• Update documentation

---

📚 Documentation preview 📚: https://RDAgent--974.org.readthedocs.build/en/974/

[Copilot]

Pull Request Overview

This PR merges the handling of proposal versions by removing v3-specific code paths and streamlining configuration and prompt instructions.

• Removed DSProposalV3ExpGen references and related conditionals from data science proposal generation.
• Updated YAML prompts for experiment submissions and feedback adjustments.
• Added support for function calling in LLM backends, including new configuration settings and abstract method definitions.

Reviewed Changes

Copilot reviewed 9 out of 11 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
rdagent/scenarios/data_science/loop.py	Removed v3-specific imports and conditional branches to unify version handling.
rdagent/scenarios/data_science/dev/prompts.yaml	Updated experiment prompt instructions to clarify submission handling.
rdagent/scenarios/data_science/dev/feedback.py	Removed untested v3-specific feedback generation logic.
rdagent/oai/llm_conf.py	Introduced a new configuration option for enabling function calling.
rdagent/oai/backend/litellm.py	Adjusted response_format handling and added support_function_calling method.
rdagent/oai/backend/deprec.py	Added support_function_calling with a FIXME comment indicating pending improvement.
rdagent/oai/backend/base.py	Added an abstract support_function_calling method to enforce subclass implementations.
rdagent/components/coder/data_science/pipeline/prompts.yaml	Removed v3-specific pipeline prompt sections.
rdagent/components/coder/data_science/pipeline/init.py	Eliminated a temporary patch for proposal version v3 in the pipeline coder.

Comments suppressed due to low confidence (2)

rdagent/scenarios/data_science/loop.py:96

• Ensure that using import_class to handle 'v3' (which is now absent from the valid list) produces the expected behavior and does not inadvertently bypass necessary initialization steps.

            if DS_RD_SETTING.proposal_version not in ["v1", "v2"]:

rdagent/oai/backend/litellm.py:93

• Review the removal of 'response_format' from kwargs, as this may affect downstream consumers expecting this field; ensure that clients are updated accordingly.

        elif not supports_response_schema(model=LITELLM_SETTINGS.chat_model) and "response_format" in kwargs: