[evaluation] Enable groundedness for Bing and OpenAPI tool calls

[mmkawale]

Summary

Enable groundedness validator support for Bing and OpenAPI tool calls while keeping stricter rejection for the remaining unsupported tool families.

Changes

• Update groundedness-specific unsupported tool list to allow:
  • bing_grounding
  • bing_custom_search
  • openapi_call
• Keep groundedness-only rejection for:
  • browser_automation
  • code_interpreter_call
  • computer_call
  • web_search
• Update groundedness validator tests to assert new allow/reject behavior.
• Refresh groundedness evaluator inline comment to match policy.

Validation

• PYTHONPATH=sdk/evaluation/azure-ai-evaluation python -m pytest sdk/evaluation/azure-ai-evaluation/tests/unittests/test_common_validators.py -k "UnsupportedToolsListGroundednessValidator or UnsupportedToolsListConversationValidator or ConversationValidatorRejectsStillUnsupportedTools"
• Result: 24 passed

Diff hygiene

This PR is a low-noise replacement for #48304 with reduced whitespace churn.

[azure-pipelines]

Azure Pipelines:
Successfully started running 1 pipeline(s).
9 pipeline(s) were filtered out due to trigger conditions.
There may be pipelines that require an authorized user to comment /azp run to run.

[Copilot]

Pull request overview

Enables groundedness validation for Bing and OpenAPI tool calls while retaining restrictions for unsupported tool families.

Changes:

• Narrows groundedness’s unsupported-tool list.
• Updates validator tests and policy comments.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

File	Description
_conversation_validator.py	Updates groundedness tool restrictions.
_groundedness.py	Refreshes evaluator policy documentation.
test_common_validators.py	Tests allowed and rejected tool families.

[mmkawale]

Addressed: updated the stale class-level policy comment in _conversation_validator.py so it no longer says Groundedness rejects structured grounding tools pending a context extractor. It now states the current policy: those structured grounding tools are accepted, and Groundedness only keeps stricter guardrails for the remaining unsupported tool families.

[Copilot]

Pull request overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated no new comments.

Comments suppressed due to low confidence (2)

sdk/evaluation/azure-ai-evaluation/azure/ai/evaluation/_evaluators/_common/_validators/_conversation_validator.py:488

• The subclass does not preserve stricter guardrails; it relaxes the base policy by removing three entries. Its purpose is to scope that relaxation to Groundedness without changing other evaluators, so the class documentation currently describes the relationship in reverse.

    A dedicated subclass is still kept so Groundedness can preserve stricter
    guardrails for the remaining unsupported tool families without changing
    behavior of other evaluators.

sdk/evaluation/azure-ai-evaluation/azure/ai/evaluation/_evaluators/_groundedness/_groundedness.py:123

• This says the groundedness policy is stricter than the shared validator, but the new list is a subset of ConversationValidator.UNSUPPORTED_TOOLS; both validators reject these four remaining families. Please avoid documenting a stricter relative policy that the code does not implement.

        # keeps stricter guardrails than the shared base for the remaining
        # unsupported tool families, while allowing grounding/search/openapi
        # tool calls supported by this evaluator.