feat: add configurable startupProbe for all dragonfly components

[snigdhasambitak]

What

Adds optional startupProbe to manager, scheduler, seedClient and client. Off by default, no behavior change unless enabled.

Why

Currently the chart only ships hardcoded readiness/liveness probes. On clusters where dfdaemon boot is slow (peer discovery, cache warm-up), the existing liveness probe can restart the container before it's actually up, or leave it NotReady well after dfdaemon is serving. startupProbe is the right fix but there's no way to set one through values today, so users end up patching rendered manifests.

Example

client:
  startupProbe:
    enable: true
    periodSeconds: 2
    failureThreshold: 90

Defaults give a 5 minute startup budget when enabled. Reuses the same grpc_health_probe exec the existing probes use, so no new image dependencies.

Chart bumped to 1.6.24, README regenerated via helm-docs.

Signed-off-by: Snigdha Aryakumar snigdha.sambit.ak@gmail.com

[snigdhasambitak]

@imeoer @gaius-qi @CormickKneey could one of you add the enhancement label so the Classify PR check passes? External contributors can't apply labels themselves. Thanks.

[gaius-qi]

LGTM