fix(i3): repair the dead idle-inhibitor autostart and make it a singleton

The i3 `exec` line pointed at
/home/kuchy/linux-configuration/scripts/utils/turn_off_auto_idle_screen_shutdown.sh
- a typo'd home ("kuchy") *and* a path that stopped existing when the repo
was reorganised in db6276b. The exec has been failing silently at every
login ever since, so the idle inhibitor was simply not running: a controller
was connected with no "idle-off" lock held, meaning the screen could blank
and lock during controller gaming, which is exactly what the line exists to
prevent.

Point it at the script's real location.

Also make --watch-controller a singleton. It is a long-lived watcher started
from an i3 `exec`, and i3 re-runs those on every `i3-msg restart`, so each
restart would leave another watcher holding a redundant inhibitor for the
rest of the session. One flock, held for the life of the watcher.

Verified: the watcher now holds
  idle-off 1000 kuhy <pid> systemd-inhibit idle "game controller connected"
and a second invocation declines instead of stacking a second inhibitor.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

docs/superpowers/evidence/i3-idle-inhibitor-path-2026-07-25.json

@@ -0,0 +1,44 @@
+{
+  "intent": "Restore the idle-inhibitor autostart that has been silently failing at every i3 login, so the screen stops blanking/locking while a game controller is connected, and stop i3 restarts from stacking duplicate watchers.",
+  "scope": [
+    "linux_configuration/scripts/periodic_background/i3-configuration/i3/config",
+    "linux_configuration/scripts/single_use/utils/turn_off_auto_idle_screen_shutdown.sh",
+    "Non-goal: no change to what the watcher does once running (the fork-free systemd-inhibit rewrite in fcd3f7e stands)",
+    "Non-goal: no i3 reload performed - a game was running on DP-0, so the live session was fixed by launching the watcher directly instead"
+  ],
+  "changes": [
+    "i3 config: the exec pointed at /home/kuchy/linux-configuration/scripts/utils/turn_off_auto_idle_screen_shutdown.sh - a typo'd home ('kuchy', which does not exist) and a path that stopped existing when the repo was reorganised in db6276b. Repointed at the script's real location under linux_configuration/scripts/single_use/utils/.",
+    "turn_off_auto_idle_screen_shutdown.sh: --watch-controller now takes an flock on $XDG_RUNTIME_DIR/idle-off-watch.lock for the life of the watcher and declines to start if one is already running. i3 re-runs its exec lines on every `i3-msg restart`, so without this each restart left another long-lived watcher holding a redundant inhibitor."
+  ],
+  "verification": [
+    {
+      "command": "systemd-inhibit --list | grep idle-off  (before the fix)",
+      "result": "fail",
+      "evidence": "No idle-off inhibitor held, while /dev/input/js0 was present - i.e. the screen could blank during controller gaming. Only rtkit/upower/xss-lock inhibitors were listed."
+    },
+    {
+      "command": "bash turn_off_auto_idle_screen_shutdown.sh --watch-controller  (after the fix)",
+      "result": "pass",
+      "evidence": "[idle-off] Holding idle inhibitor (pid 941016) while a controller is connected; systemd-inhibit --list now shows: idle-off 1000 kuhy 941016 systemd-inhibit idle 'game controller connected' block"
+    },
+    {
+      "command": "second invocation of the same script (singleton check)",
+      "result": "pass",
+      "evidence": "[idle-off] Controller watcher already running; not starting a second one. Exit 0, and `systemd-inhibit --list | grep -c idle-off` stays at 1."
+    },
+    {
+      "command": "pre-commit run --files linux_configuration/scripts/single_use/utils/turn_off_auto_idle_screen_shutdown.sh",
+      "result": "pass",
+      "evidence": "All hooks Passed or Skipped, including shellcheck and 'Block polling script anti-patterns'."
+    }
+  ],
+  "risks": [
+    "The lock lives in $XDG_RUNTIME_DIR, which is cleared on logout - correct here, since the watcher is per-session anyway. If XDG_RUNTIME_DIR is unset the lock falls back to /tmp, where a stale lock file is harmless because flock releases on process exit.",
+    "The i3 config change only takes effect on the next i3 restart/login. The running session was fixed by starting the watcher directly, so there is no gap; a future `i3-msg restart` will pick up the corrected path and the singleton guard will prevent a duplicate.",
+    "The deployed ~/.config/i3/config is a separate copy, not a symlink to this tracked file. Both were updated; if only one had been, the next deploy would have reintroduced the broken path."
+  ],
+  "rollback": [
+    "git revert the commit, then restore the deployed copy: the pre-change ~/.config/i3/config is backed up in the session scratchpad as i3-config.bak.",
+    "After rollback, confirm with `systemd-inhibit --list | grep idle-off` that the inhibitor state matches expectations, and kill any running turn_off_auto_idle_screen_shutdown.sh watcher."
+  ]
+}

linux_configuration/scripts/periodic_background/i3-configuration/i3/config

@@ -35,7 +35,7 @@ exec --no-startup-id xss-lock --transfer-sleep-lock -- i3lock --nofork
 exec --no-startup-id nm-applet
 
 # Keep screen awake and unlocked; also treat controller input as activity
-exec --no-startup-id bash /home/kuchy/linux-configuration/scripts/utils/turn_off_auto_idle_screen_shutdown.sh --watch-controller
+exec --no-startup-id bash /home/kuhy/testsAndMisc/linux_configuration/scripts/single_use/utils/turn_off_auto_idle_screen_shutdown.sh --watch-controller
 
 # Use pactl to adjust volume in PulseAudio.
 set $refresh_i3status killall -SIGUSR1 i3status

linux_configuration/scripts/single_use/utils/turn_off_auto_idle_screen_shutdown.sh

@@ -270,6 +270,14 @@ main() {
   persist_with_systemd_logind
 
   if [[ $watch_controller == true ]]; then
+    # Singleton. i3 re-execs its `exec` lines on `i3-msg restart`, and a second
+    # long-lived watcher would sit there holding a redundant inhibitor for the
+    # rest of the session. One lock, taken for as long as the watcher lives.
+    exec 8> "${XDG_RUNTIME_DIR:-/tmp}/idle-off-watch.lock"
+    if ! flock -n 8; then
+      log "Controller watcher already running; not starting a second one."
+      exit 0
+    fi
     log "Controller activity watcher enabled (idle-inhibitor mode)"
     # Blocks until terminated; releases the inhibitor on exit via its own trap.
     start_controller_watchers