feat(ci): add GC CronJob for runner cleanup, expand reg PVCs to 20Gi

- Add gc-cronjob.yaml: daily prune of DinD Docker images/volumes/build-cache
  and actcache across all forgejo-runner pods. Keeps :latest tagged images,
  deletes non-latest older than 72h.
- Expand runner reg PVCs from 1Gi to 20Gi (all three runners) to prevent
  action tool cache from filling disk.
- Rust runner gets explicit 20Gi persistence override.
- GC only renders from golang (default) values to avoid duplicate resources.

k8s/infra/forgejo-runner/values.yaml

@@ -39,7 +39,7 @@ dind:
 persistence:
   reg:
     storageClass: longhorn  # Unified StorageClass (3 replicas)
-    size: 1Gi               # .runner registration file + config — survives pod restarts
+    size: 20Gi              # .runner registration file + action tool cache + actcache artifacts
   dind:
     storageClass: longhorn  # Unified StorageClass (3 replicas)
     size: 30Gi              # docker layer cache — keeps rebuilds fast across restarts
@@ -53,3 +53,14 @@ tolerations:
 # attach there.
 nodeSelector:
   topology.kubernetes.io/zone: az-a
+
+# GC CronJob — prunes Docker images/volumes/build-cache and actcache across
+# ALL forgejo-runner pods. Only enable in default values (golang instance);
+# disable in per-runner overrides so it renders once.
+gc:
+  enabled: true
+  schedule: "0 3 * * *"        # daily 03:00 UTC
+  image: alpine/k8s:1.31.0
+  pruneAge: "72h"              # Docker artifacts unused longer than this get pruned
+  pruneAgeHours: 72             # Same as pruneAge but numeric for date arithmetic in shell
+  actcacheMaxAgeDays: 3        # actcache files older than N days get deleted