Redesign Disks section: group datasets under a shared pool header

Problem

The current Disks UI shows each dataset as an independent card with identical pool-level storage numbers (size, used, free, progress bar). In a shared ZFS pool with multiple datasets, this is misleading — it looks like each dataset has its own independent storage, when in reality they all share the same pool.

Solution

• Add a pool header showing the true shared storage stats: total size, progress bar, used size + percentage, free percentage
• Below the header, render each dataset as its own card with dataset-specific info: mode, clones, snapshots, size
• When a pool has no child datasets (e.g. dblab_pool_01), suppress the redundant dataset name since the pool header already identifies it

Acceptance criteria

• Single-pool setup shows pool header + single card without repeated name
• Multi-dataset setup groups datasets under one pool header
• Storage stats on pool header match actual ZFS pool usage
• >80% disk usage warning displays correctly (only when at least one dataset is active)
• Old fileSystem API fallback still works
• Empty pools array doesn't render a blank section