## Symptom
On an HP Victus 15-fa1xxx (Intel Raptor Lake-P iGPU + NVIDIA RTX 4050 Mobile, BOE eDP panel), closing the lid suspends fine, but opening it does not resume instantly:
1. Lock screen appears, but keyboard (AT Translated Set 2) and touchpad (ELAN0788 on i2c) accept no input for a few seconds.
2. Screen goes black, then the lock screen comes back and input works.

## Evidence (local journal, redacted)
Lock itself succeeds fast before suspend, so this is not a lock-service failure:
```
22:23:17 systemd-logind: Lid closed / Suspending...
22:23:17 omarchy-shell: lock-requested / lock-pending: screen-stabilizing
22:23:17 omarchy-shell: secure=true
22:23:17 systemd-sleep: Performing sleep operation 'suspend' (s2idle)
22:23:44 systemd-logind: Lid opened.
22:23:45 systemd-sleep: System returned from sleep operation 'suspend'.
```
The display driver fails panel link training on resume — and keeps failing afterwards:
```
i915 0000:00:02.0: [drm] *ERROR* [CONNECTOR:508:eDP-1][ENCODER:507:DDI A/PHY A][DPRX] Failed to enable link training
```
Seen at 22:23:44 (resume), then again 22:24:01, 22:24:41, 22:26:45 with no further suspends. Also on resume: `i915 raw-wakerefs=1 wakelocks=1 on cleanup` warning and an nvidia-drm flip-event warning. `/sys/power/mem_sleep` offers only `[s2idle]` (no deep S3), kernel `7.2.5-3-omarchy`, Hyprland 0.56.2, BIOS F.18.

Related but distinct from #8112 (sleep-lock inhibitor race): that failure line also appears here (`Failed to inhibit: already running` on the respawned monitor), but the lock had already gone secure, so it does not explain the flicker/dead input.

## Likely contributors in Omarchy defaults/user config
- `~/.config/hypr/monitors.lua` forced `eDP-1` to `1920x1080@144` (panel max) and forced `HDMI-A-1` to `1920x1080@60` at a fixed position even with nothing plugged in (`hyprctl monitors all -j` shows only eDP-1). Every resume therefore does a 144Hz eDP modeset known-flaky with PSR on this iGPU, plus a dead-output probe.
- The shell holds the session lock 500ms for screen stabilization (`lock/Service.qml` `sessionLockStabilizeTimer`), which is correct, but the password field is presented before the i2c input devices finish re-probing, so first-seconds typing goes nowhere.

## Local workaround applied (safe, no reboot)
- `eDP-1` → `1920x1080@60`, `HDMI-A-1` → `mode preferred, position auto` so resume is a single clean 60Hz modeset. `hyprctl configerrors` clean, `hyprctl monitors` confirms 60Hz. Still to test across several lid cycles; kernel option `i915.enable_psr=0` not yet tried (needs Limine cmdline + reboot).

## Suggestions for Omarchy
1. Do not ship/encourage a forced HDMI rule when no external monitor is connected — prefer `preferred/auto` or commented-out, so resume does one eDP modeset.
2. Consider calling out 144Hz + PSR resume flakiness (or a 60Hz fallback note) for Raptor Lake-P BOE panels.
3. Consider gating password-field focus/accept on input-device readiness after resume, or showing a brief 'waking devices…' state, since the lock currently looks ready while i2c input is still down.