fix: try watchtower image that respects container timeouts

[skylenet]

Based on https://github.com/skylenet/watchtower/tree/auto-detect-stop-timeout

I've left the 290 on purpose. Whenever a container gets updated, we should see a timeout of 300s (which is what we use for most containers). The 290s should only appear if a container doesn't have a defined Timeout.

[skylenet]

This is confirmed to be working. It respected the 300s from the container.

time="2026-01-12T07:02:55Z" level=info msg="Found new image" container=execution image="docker.ethquokkaops.io/dh/ethpandaops/geth:master" new_id=f550c99bcfb2
time="2026-01-12T07:02:56Z" level=info msg="Stopping container" container=execution id=b3222553cec9 signal=SIGTERM timeout=5m0s
time="2026-01-12T07:04:02Z" level=info msg="Started new container" container=execution id=b3222553cec9 new_id=04dd974ccdd2