Fix screen flash when waking from sleep into the unlock screen

The lock service arms a 5s blank timer whenever the screen locks, and
input at the lock screen re-arms it. Closing the lid sprays pointer
noise over the lock surface, so the timer was routinely armed right
before suspend, froze mid-countdown, and fired moments after resume --
blanking the freshly woken unlock screen under the user.

Guard the timer with a wall-clock check: if far more time elapsed than
the interval, the countdown slept through a suspend, so take a fresh
run-up instead of blanking. This also blanks the lock screen 5s after
an untouched resume.

Two accomplices made the flash worse and hid the real bug:

- The clamshell watcher's 2s poll fired an unconditional global DPMS
  enable whenever no external monitor was active, relighting any blank
  within 2 seconds (lock-screen blanking never stuck on undocked
  laptops) and racing the resume modeset. Recovery now only wakes
  displays when it actually re-enables one.

- Every keystroke at the lock screen dispatched a redundant DPMS
  enable via omarchy-system-wake, forcing extra modesets in the
  fragile just-resumed DRM state. Brightness "on" now skips the
  dispatch when every active display is already lit.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

test/shell.d/monitor-recovery-test.sh

@@ -58,7 +58,9 @@ grep -F "hyprctl dispatch 'hl.dsp.dpms({ action = \"enable\" })' >/dev/null 2>&1
 grep -F 'hyprctl monitors all -j' "$monitor_internal" >/dev/null
 grep -F 'omarchy-hyprland-monitor-external-active' "$monitor_internal" >/dev/null
 grep -F 'wake' "$monitor_internal" >/dev/null
+grep -F 'omarchy-hyprland-toggle-enabled $TOGGLE || return 0' "$monitor_internal" >/dev/null
 pass "internal monitor helper can re-enable disabled laptop displays"
+pass "internal monitor recovery only wakes displays when it re-enables one"
 
 grep -F 'omarchy-hyprland-monitor-external-active' "$monitor_mirror" >/dev/null
 pass "internal mirror helper recovers when no active external display remains"