fix: filter temp: state keys from sub-agent delta in AgentTool

[nuthalapativarun]

Please ensure you have read the contribution guide before creating a pull request.

Link to Issue or Description of Change

Problem:
AgentTool.runAsync() propagates every key from a sub-agent's stateDelta to the parent context via toolContext.state.update(event.actions.stateDelta). Keys prefixed with State.TEMP_PREFIX (temp:) are explicitly designed to be ephemeral, per-invocation state that must not be persisted or cross agent boundaries. Because no filtering was applied, any temp: key written by the sub-agent leaked into the parent session state, overwriting or polluting it.

Solution:
Before calling toolContext.state.update(), filter out all keys that start with State.TEMP_PREFIX. If the resulting delta is empty (all keys were temp:), skip the update entirely.

This aligns with the same filtering applied in BaseSessionService.trimTempDeltaState() at session-persistence time, and matches the behaviour of the Python ADK.

Testing Plan

Unit Tests:

• I have added or updated unit tests for my change.
• All unit tests pass locally.

Summary of passed npm test results:

✓ |unit:core| core/test/tools/agent_tool_test.ts (3 tests)
✓ |integration| tests/integration/tools/agent_tool_test.ts (1 test)

Test Files  2 passed (2)
     Tests  4 passed (4)

Full suite: 1120 passed | 21 skipped (1141)

Manual End-to-End (E2E) Tests:
The filtering is applied only to keys that start with temp:. Non-temp state delta propagation is unchanged and verified by the existing integration test.

Checklist

• I have read the CONTRIBUTING.md document.
• I have performed a self-review of my own code.
• I have commented my code, particularly in hard-to-understand areas.
• I have added tests that prove my fix is effective or that my feature works.
• New and existing unit tests pass locally with my changes.
• I have manually tested my changes end-to-end.
• Any dependent changes have been merged and published in downstream modules.

[nuthalapativarun]

Hi team, gentle ping on this one — happy to address any feedback or make adjustments if needed. Thanks for reviewing!

[nuthalapativarun]

Hi team, gentle ping on this one — happy to address any feedback or make adjustments if needed. Thanks for reviewing!