### What happens

`omarchy-drive-info` drops the vendor/model for any drive that has a nested
device-mapper layer (LUKS, LVM, MD RAID) somewhere under it. On this machine
`/dev/sda` is a Samsung SSD 870 EVO 1TB with an encrypted root:

```
$ omarchy-drive-info /dev/sda
/dev/sda (931.5G) [crypto_LUKS]              # no model

$ omarchy-drive-info /dev/sda2
/dev/sda2 (929.5G) [crypto_LUKS]             # no model
```

Expected (and what `lsblk` reports fine on its own):

```
/dev/sda (931.5G) - ATA Samsung SSD 870 EVO 1TB [vfat(/boot), crypto_LUKS]
```

Unencrypted drives on the same machine are labelled correctly, which is what
makes this look like "that drive just has no model" rather than a bug.

### Cause

`bin/omarchy-drive-info` resolves the parent disk with:

```bash
root_drive=$(lsblk -no PKNAME "$drive" 2>/dev/null | tail -n1)
```

`lsblk -no PKNAME <dev>` prints one line for the device *and every descendant*,
so `tail -n1` returns the **deepest** parent in the subtree rather than the
parent of the device that was asked about:

```
$ lsblk -no PKNAME /dev/sda
          <- sda itself, no parent
sda       <- sda1
sda       <- sda2
sda2      <- /dev/mapper/root, the LUKS mapping
```

`tail -n1` yields `sda2`, so the script then queries a *partition* for
VENDOR/MODEL, and partitions report neither. The extra depth from the crypt
mapping is the whole difference: drives without one end the listing on the disk
itself and happen to work.

```
$ lsblk -no PKNAME /dev/sda | tail -n1
sda2                                          # a partition
```

### Why it matters

`omarchy-drive-select` builds its picker from `omarchy-drive-info`, and
`iso2sd` (in `default/bash/fns/drives`) uses that picker to choose the target
for `sudo dd ... of="$drive"`. An encrypted drive shows up in that list with no
make or model — only a size — while writing an image to the wrong device is
destructive and unrecoverable. `omarchy-drive-password` selects drives the same
way.

### Steps to reproduce

1. Have a disk with a LUKS (or LVM/MD) layer on one of its partitions.
2. Run `omarchy-drive-info /dev/<that disk>`.
3. The `- <vendor> <model>` segment is missing; `lsblk -dno MODEL /dev/<disk>`
   shows the model is available.

### Suggested fix

Take the first non-empty line instead of the last:

```diff
-root_drive=$(lsblk -no PKNAME "$drive" 2>/dev/null | tail -n1)
+root_drive=$(lsblk -no PKNAME "$drive" 2>/dev/null | awk 'NF{print; exit}')
```

The first non-empty PKNAME is the immediate parent of the queried device, which
is what the surrounding comment ("Find the root drive in case we are looking at
partitions") intends. Verified against every block device on this machine:

| argument | current | with fix |
|---|---|---|
| `/dev/sda` (disk, LUKS below) | model missing | **model restored** |
| `/dev/sda2` (the LUKS partition) | model missing | **model restored** |
| `/dev/sda1` (plain partition) | correct | unchanged |
| `/dev/sdb` (disk, no crypt) | correct | unchanged |
| `/dev/nvme0n1` (disk, no crypt) | correct | unchanged |
| `/dev/nvme0n1p3` (partition) | correct | unchanged |

Two cases fixed, four byte-identical, no regressions.

Happy to open a PR if that is useful.

### System

- Omarchy 4.0.0-1
- Kernel 7.1.8-arch1-3
- Hyprland v0.56.2
- AMD Ryzen 7 5800XT / Radeon RX 9060 XT (Navi 44)
- util-linux 2.42.2 (`lsblk`)

The affected line is unchanged on `master` as of filing.