Inconsistently the run_scenario job runs far longer than expected

[mattyg]

Inconsistently I've seen run_scenario nomad jobs that seem to run for much longer than expected.

For example, this remote_signals 5 minute job ran for over an hour: https://nomad-server-01.holochain.org:4646/ui/jobs/remote_signals@default/allocations?activeTask=c19b37b4-d6a1-2a9c-e9e0-58d726daa0f6-run_scenario&status=%5B%22running%22%5D&version=%5B2%5D

It was triggered by this CI run: https://github.com/holochain/wind-tunnel/actions/runs/23456025822/job/68250721580