Complete disk collector rewrite with clean architecture

Replaced complex disk collector with simple lsblk → df → group workflow.
Supports both physical drives and mergerfs pools with unified metrics.
Eliminates configuration complexity through pure auto-discovery.

- Clean discovery pipeline using lsblk and df commands
- Physical drive grouping with filesystem children
- MergerFS pool detection with parity heuristics
- Unified metric generation for consistent dashboard display
- SMART data collection for temperature, wear, and health

CLAUDE.md

@@ -229,34 +229,44 @@ Storage:
 - Enhanced pool health calculation (healthy/degraded/critical)
 - Hierarchical tree visualization with data/parity disk separation
 
-**🔄 In Progress - Unified Pool Visualization:**
+**🔄 In Progress - Complete Disk Collector Rewrite:**
 
-Current auto-discovery works but displays filesystems separately instead of grouped by physical drives. Need to implement unified pool concept where single drives are treated as pools.
+The current disk collector has grown complex with mixed legacy/auto-discovery approaches. Planning complete rewrite with clean, simple workflow supporting both physical drives and mergerfs pools.
 
-**Current Display (needs improvement):**
-```
-● /boot: (separate entry)
-● /nix_store: (separate entry) 
-● /: (separate entry)
-```
+**New Clean Architecture:**
 
-**Target Display (unified pools):**
+**Discovery Workflow:**
+1. **`lsblk`** to detect all mount points and backing devices
+2. **`df`** to get filesystem usage for each mount point
+3. **Group by physical drive** (nvme0n1, sda, etc.)
+4. **Parse `/proc/mounts`** for mergerfs pools
+5. **Generate unified metrics** for both storage types
+
+**Physical Drive Display:**
 ```
 ● nvme0n1:
-  ├─ Drive: T: 35°C W: 1%
-  ├─ /boot: 11% 0.1GB/1.0GB
-  ├─ /nix_store: 23% 214.9GB/928.2GB
-  └─ /: 23% 214.9GB/928.2GB
+  ├─ ● Drive: T: 35°C W: 1%
+  ├─ ● Total: 23% 218.0GB/928.2GB
+  ├─ ● /boot: 11% 0.1GB/1.0GB
+  └─ ● /: 23% 214.9GB/928.2GB
 ```
 
-**Required Changes:**
-1. **Enhanced Auto-Discovery**: Group filesystems by backing physical drive during discovery
-2. **UI Pool Logic**: Treat single drives as "pools" with drive name as header
-3. **Drive Info Display**: Show temperature, wear, health at pool level for single drives
-4. **Filesystem Children**: Display mount points as children under their physical drives
-5. **Hybrid Rendering**: Physical grouping for single drives, logical grouping for mergerfs pools
+**MergerFS Pool Display:**
+```
+● /srv/media (mergerfs):
+  ├─ ● Pool: 63% 2355.2GB/3686.4GB
+  ├─ Data Disks:
+  │  ├─ ● sdb T: 24°C
+  │  └─ ● sdd T: 27°C  
+  └─ ● sdc T: 24°C (parity)
+```
 
-**Expected Result**: Consistent hierarchical storage visualization where everything follows pool->children pattern, regardless of underlying storage technology.
+**Implementation Benefits:**
+- **Pure auto-discovery**: No configuration needed
+- **Clean code paths**: Single workflow for all storage types
+- **Consistent display**: Status icons on every line, no redundant text
+- **Simple pipeline**: lsblk → df → group → metrics
+- **Support for both**: Physical drives and mergerfs pools
 
 ## Important Communication Guidelines