When set, renders `startupProbe` on the runner container with the
operator-supplied spec. While the startupProbe is running kubelet
does not run the existing exec-based liveness / readiness probes,
so the pod stays NotReady — and during a rolling update the old
pod is not terminated — until the startup probe passes.

Motivation: deeper boot-time health checks (e.g. dispatching a
synthetic job through the standard Build pipeline before accepting
real jobs) need a way to gate readiness without overriding the
existing pgrep-based readinessProbe, which would lose its
runtime semantic of "is the runner process still alive".

The block defaults to empty so the rendered Deployment is
byte-identical for any operator who doesn't opt in.

Add optional startupProbe to the runner Deployment

See merge request !551

Merged-by: Igor <iwiedler@gitlab.com>
Approved-by: Tomasz Maczukin <tomasz@maczukin.pl>
Reviewed-by: Tomasz Maczukin <tomasz@maczukin.pl>

Rendered only when .Values.podDisruptionBudget is set, so existing
deployments see no change.

Add optional PodDisruptionBudget for the runner manager Deployment

See merge request !553

Merged-by: Igor <iwiedler@gitlab.com>
Approved-by: Tomasz Maczukin <tomasz@maczukin.pl>