### What happened

With a 4-pole headset already plugged in at boot, the ALC245 codec comes up in
the "plain headphones" configuration: the `Headset Mic Jack` control reads
`off`, and headphone playback is muffled/attenuated. Unplugging and re-inserting
the plug — or re-probing the PCI audio device — makes the codec report
`Headset Mic Jack = on` and playback is correct.

### What was expected

The headset to be detected as a headset at boot, as it is on the stock Arch
kernel on the same machine.

### A/B, same machine, same userspace, same UCM

| kernel | `Headset Mic Jack` at boot | playback |
|---|---|---|
| `linux 7.2.3-arch1-3` (stock Arch, also installed) | `on` | correct, no replug needed |
| `linux-omarchy 7.2.5-3` (default since 4.0.4) | `off` | muffled until replug |

Nothing in userspace differs between the two boots: same alsa-ucm-conf
1.2.16.1-1, same sof-firmware 2025.12.2-1, same PipeWire/WirePlumber state.

### Steps to reproduce

1. Boot `linux-omarchy` with the 4-pole headset already plugged in.
2. In that state:
       amixer -c1 cget iface=CARD,name='Headphone Jack'    -> values=on
       amixer -c1 cget iface=CARD,name='Headset Mic Jack'  -> values=off
   Headphone playback is muffled.
3. Unplug and re-insert the plug:
       amixer -c1 cget iface=CARD,name='Headset Mic Jack'  -> values=on
   Playback is correct.

### Evidence that it is not userspace configuration

Everything readable from userspace is already correct while the audio is muffled:

- active card profile: `HiFi (HDMI1, HDMI2, HDMI3, Headphones, Mic1, Mic2)`
- codec node 0x21: `Pin-ctls: 0xc0: OUT HP` (already HP out, not line out),
  `Unsolicited: tag=01, enabled=1`
- DAC node 0x03, the selected connection of 0x21: `Amp-Out vals: [0x54 0x54]`,
  i.e. the 84/87 that `Headphone Playback Volume` reports
- speaker pin 0x14 is `off`; setting `Auto-Mute Mode` to `Enabled` changes nothing
- PipeWire default sink 90%, unmuted; Vivaldi and Chromium streams unmuted

`Headset Mic Jack` is `access=r-------`, so userspace cannot set it, and no UCM
device/verb for this card is conditioned on that control.

Re-probing the PCI audio device also fixes it without touching the plug, so the
misdetection is specific to the cold boot probe path:

    echo 0000:00:1f.3 > /sys/bus/pci/drivers/sof-audio-pci-intel-tgl/unbind
    sleep 3
    echo 0000:00:1f.3 > /sys/bus/pci/drivers/sof-audio-pci-intel-tgl/bind

### System

- Omarchy 4.0.4-1
- Acer Predator PHN16-72, BIOS V1.11
- Realtek ALC245 (subsystem id 0x10251731); PCI 00:1f.3 Intel 700 Series Chipset
  Family HD Audio [8086:7a50] (rev 11)
- SOF: `sof-audio-pci-intel-tgl`; machine driver `skl_hda_dsp_generic`; topology
  `intel/sof-tplg/sof-hda-generic-4ch.tplg`
- alsa-utils 1.2.16-1, alsa-ucm-conf 1.2.16.1-1, sof-firmware 2025.12.2-1
- diagnostics: `omarchy-debug` output and `alsa-info.sh --no-upload` output are
  available on request (gh cannot attach files to an issue)

### Related

Not the same failure, but possibly relevant — other regressions introduced in the
same kernel build, or the same "differs from stock Arch" pattern:

- #12238 i915 CPU pipe A FIFO underrun since linux-omarchy 7.2.5-3
- #12188 linux-omarchy kernel breaks screen backlight and USB audio volume on Dell XPS 14
- #12149 linux-omarchy 7.2.5-3 ships with binder disabled (stock Arch `linux` enables it)

### Workaround

Boot the stock `linux` kernel, which is not affected.