Runner machine metrics now populate in the CI job dashboard

The Metrics tab for Tuist Runners CI jobs now shows live machine data (CPU, memory, network, storage, and I/O wait) for every job, including the summary charts on the Overview page. The previous metrics collector ran in the runners-controller and failed silently in production, so all charts were empty. Sampling has moved inside the runner VMs themselves: macOS runners report whole-VM metrics via a background script, and Linux runners report via a dedicated metrics sidecar. The server now accepts these samples directly from the runner Pod, and a leaked token cannot write another Pod’s metrics.