HPA & KEDA Autoscaling

Autoscaling gets first-class treatment: HPA views are easier to read at a glance, and KEDA joins as Zapkube's second built-in plugin — alongside CRD printer columns, richer copy tools, and a native macOS menu bar.

📈 HPA & KEDA

• KEDA plugin — ScaledObjects are now a first-class, built-in plugin: their triggers, current metrics, and the HPA they generate are all shown together, without needing a separate lookup.
• Clearer HPA overview — gauges are capped to a sensible column width, every metric fits in one tooltip instead of being cut off, and each metric row gets its own line with the qualifier separated from the bar.
• Autoscaling section — the old "autoscaled" badge on workload overviews is now a proper Autoscaling section showing which controller owns the replica count.
• Scale dialog warnings — manually scaling a workload now warns when an HPA (including one generated by KEDA) already controls its replica count, even through inherited ownership like a ReplicaSet.

🧩 CRDs

• Printer columns — Custom Resource lists now read additionalPrinterColumns straight from the CRD and render them as regular sortable, filterable table columns, even for CRDs with no dedicated module.

📋 Copy & Permissions

• Copy modifiers everywhere — click to copy a value, ⌥-click for NAME=value, on env vars, labels, and annotations; copy buttons on masked Secrets are marked so you know what you're copying before you paste it.
• ServiceAccount permissions — the ServiceAccount page now lists every RBAC rule that applies to it, not just a summary.
• YAML editor — long manifests get collapsible sections and sticky headers so you always know which block you're looking at.

🍎 macOS

• Native menu bar, with a new setting for whether quitting the app closes the last window or keeps it running in the background with active port-forwards intact.

🐛 Fixes

• Pods with dozens of container ports no longer overflow their card layout.
• Port-forwarding no longer offers a forward on a UDP port.

🧹 Misc / Chore

• Table rendering is noticeably faster, especially on large resource lists.
• Dependency updates across the frontend and Go toolchain.