fix(analytics): log error-bearing tool results as TOOL_ERROR

[Zsanz3]

Fixes #7112

Summary

BigQueryAgentAnalyticsPlugin.after_tool_callback was logging error-bearing tool results as TOOL_COMPLETED / OK. It now classifies results with isError / error_details as TOOL_ERROR (aligned with on_tool_error_callback).

Changes

• Detect error_details, isError / is_error; prefer MCP content[0].text for error_message when present
• Focused unit test with mocked _log_event covering MCP isError and error_details
• Diff kept tiny; no ReflectAndRetry changes

Tests

pytest tests/unittests/plugins/test_bigquery_agent_analytics_plugin.py -q

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.

[paulxg12]

Thanks for addressing error-bearing results that would otherwise be logged as successful tool calls. The new error_message is None guard on final-response tools caught my attention too: it prevents a failed call from also producing an AGENT_RESPONSE. Is that case covered by an existing test? If not, a case with the tool in final_response_tool_names and an error-bearing result would help protect that behavior. This is from reading the diff; I haven't run the suite locally.

[Zsanz3]

Thanks @paulxg12 — good catch on the error_message is None guard.

I added a unit test that puts the tool in final_response_tool_names, returns an error-bearing result (error_details), and asserts we get TOOL_ERROR without an AGENT_RESPONSE. Appreciate you reading the diff carefully.