Clamp wait_for_agents timeout to the tool's ceiling

[bearsyankees]

Models sometimes request very large wait timeouts (hours or more). The raw value went straight into the inner asyncio.wait_for, so the SDK's tool-call ceiling fired first and aborted the call instead of returning the clean timeout payload that tells the agent to keep working or finish. Clamp the requested value into [1s, 300s], wait on the clamped value, and echo it back.

[greptile-apps]

Greptile Summary

This PR clamps wait_for_agents timeouts to the supported 1–300 second range so the inner timeout can return a clean payload before the SDK ceiling fires.

• Adds centralized timeout normalization and reports the effective timeout in timeout responses.
• Adds coverage for oversized, non-positive, in-range, and early-return cases.

Confidence Score: 5/5

The PR appears safe to merge with no concrete correctness or security defects identified.

The timeout is normalized before reaching asyncio.wait_for, the returned payload consistently reports the effective duration, and tests cover the affected timeout boundaries and early-return paths.

Important Files Changed

Filename	Overview
strix/tools/agents_graph/tools.py	Clamps requested wait durations before the inner wait and reports the effective value without changing early-return behavior.
tests/test_wait_dedupe.py	Adds focused tests for timeout clamping, payload reporting, and existing deduplication paths.

_{Reviews (1): Last reviewed commit: "Clamp wait_for_agents timeout to the too..." | Re-trigger Greptile}