Bare-metal fleet MachineHealthChecks in the Helm chart allow the full OS reinstall window

Self-hosting operators who deploy the Tuist Helm chart on bare-metal providers (Scaleway Dedibox, OVHcloud, Scaleway Elastic Metal) can now run helm upgrade against canary or production without the chart’s fleet MachineHealthChecks trapping still-installing boxes in an OS reinstall loop. The dedibox, ovh, and kura fleet MHCs now set nodeStartupTimeout to 90m (up from 30m), giving a box the full reinstall + boot window to join the cluster instead of being marked unhealthy every 30 minutes, wiped, reinstalled, and restarted at zero. Steadystate remediation for boxes that fail after joining remains at 10 minutes via unhealthyConditions, so the longer startup window only applies to the initial provisioning / reinstall phase.