MacOS runner cache is now reachable after a Private Network DHCP lease failure

Fixed an issue where runner-cache Elastic Metal nodes could silently lose their Scaleway Private Network address when the DHCP lease could not be renewed. When that happened, the node’s module-cache NodePort became unreachable, causing macOS CI builds to hang at “Fetching remote binaries via module cache. Hold tight…” with repeated timeouts. The node now discovers its Private Network address via DHCP once and then holds it as a static address, so the cache stays reachable and builds can continue to fetch remote binaries reliably.