e2e(gateway): parent-child-and-route-drive-listener-set caps the operation at 6m below its own waits #4114

Describe the bug
hack/e2e-chainsaw/gateway/chainsaw-test.yaml, step parent-child-and-route-drive-listener-set (line 829 on main): the operation carries timeout: 6m (line 832) while the waits inside its script add up to more than that (180+120+180+120 seconds and further waits down the script). When the inner waits run to their limits, chainsaw kills the process group at the six-minute mark, the cleanup trap never runs, and the tenant namespaces the step created stay in front of every later suite. .chainsaw.yaml:53 and docs/agents/e2e-testing.md:127 both state the rule the step breaks: per-wait budgets must sum to less than the operation ceiling.

Environment

• Cozystack main, 2026-09-07

To Reproduce

1. Read lines 829-900 of the file and add up the timeout N prefixes against the timeout: 6m above them

Expected behaviour
The operation ceiling exceeds the sum of the inner waits, so a slow step fails red with its diagnostics and its cleanup, as the sibling steps do.

Actual behaviour
A slow step is killed from outside, with no cleanup.

Additional context
The step added by #3342 next to it was sized the other way round (per-wait budgets under an explicit ceiling) after the same review caught the same shape there.

Checklist

• I have checked the documentation
• I have searched for similar issues
• I have included all required information
• I have provided clear steps to reproduce
• I have included relevant logs