Add optional adaptive escalation verification for Plan-Execute

[HemVadgama]

Summary

This PR introduces an experimental, flag-gated adaptive escalation mechanism for the Plan-Execute runner.

The goal is to provide a small, incremental foundation for adaptive orchestration without changing the default execution behavior.

Changes

• Added deterministic escalation policy based on execution signals.

• Added optional adaptive_escalation flag to PlanExecuteRunner.

• Default behavior remains unchanged (adaptive_escalation=False).

• When enabled:

  • execution signals are extracted after plan execution,
  • observability attributes are recorded,
  • an additional verification pass is performed only if the deterministic policy decides escalation is necessary,
  • verification notes are incorporated into the final summarization.

Testing

Verified with:

uv run pytest src/agent/tests/test_escalation.py
uv run pytest src/agent/tests/test_runner.py
uv run pytest src/agent/tests

All tests pass.

Notes

This is intended as an incremental step toward evaluating adaptive orchestration strategies rather than a replacement for the existing Plan-Execute architecture.

Related to #360

[DhavalRepo18]

@ShuxinLin What is plan for this PR?

[DhavalRepo18]

Hi Hem,

We will review this feature, but you need to provide a technical report on why escalation is important.

What benefits does it bring?