test: prune flaky/failing functional tests and trim CI matrix

[hadi-aix]

Summary

Cleans up functional tests that are currently failing against the test backend and trims the CI matrix accordingly. Changes are grouped as removed, skipped, and CI.

Removed tests

• tests/functional/model/run_utility_model_test.py — deleted the six utility-model run tests that fail with ResponseStatus.FAILED instead of SUCCESS: test_run_utility_model, test_utility_model_with_decorator, test_utility_model_string_concatenation, test_utility_model_code_as_string, test_utility_model_simple_function, test_utility_model_update. Kept test_utility_model_status and test_model_tool_creation; dropped the now-unused UtilityModelInput/utility_tool import.
• Metrics tests — removed all metrics-related tests:
  • tests/functional/general_assets/asset_functional_test.py: test_list_metrics and test_run_metric (plus the unused MetricFactory import).
  • tests/functional/pipelines/designer_test.py: test_metric_pipeline (metric node) and its now-unused DatasetFactory import.

Skipped tests (@pytest.mark.skip with reason)

• tests/functional/model/run_model_test.py — test_index_model_with_filter (all four embedding-model params: OpenAI Ada 002, Jina Clip v2 Multimodal, Multilingual E5 Large, BGE M3) and test_index_model_air_with_image (sample image asset 404s during document parsing).
• tests/functional/model/run_connect_model_test.py — test_run_script_connection_tool and test_run_script_connection_tool_with_complex_inputs (script tool run returns Failure).
• tests/functional/v2/test_agent.py — test_slack_tool_integration_with_agent (backend rejects the Slack send-message payload).
• tests/functional/agent/ and tests/functional/team_agent/ — new directory-scoped conftest.py files skip every test collected under each directory (path-scoped so running the whole suite doesn't skip unrelated tests). Tests still show as skipped rather than silently uncollected.

CI

• .github/workflows/main.yaml — removed the benchmark, pipeline_3.0_v1, pipeline_designer, and pipeline_create steps from the test matrix (both the test-suite list and the include block).

Notes

• "Remove all tests related to metrics" was interpreted to include test_metric_pipeline (a metric-node pipeline test). The benchmark suite uses metrics but is about benchmarking, so it was left as-is (and is dropped from CI anyway).
• All edited Python files pass py_compile.

🤖 Generated with Claude Code