Docs: public copy, site presentation, release-gate review (D533, D534)

Public-copy language pass, homepage restructure and Documentation rename,
404 page, dead-anchor and permalink fixes, release-gate sign-off, and the
pre-release review fixes of the 2026-10-05 batch (probe-doc drift,
build-path hygiene, verify.sh and capture safety).

Refs: D533 D534

tools_pc/audiodebug.ps1

@@ -48,7 +48,7 @@ param(
     [switch]$Dump,                  # GE_AUDIODUMP=1  -> audiodump.raw
     [switch]$VoiceDump,             # GE_VOICEDUMP=1  -> voicedump.raw (per-voice, D202/M-67)
     [switch]$MixerTrace,            # GE_MIXERTRACE=1 -- SLOW, see warning below
-    # (the -AB/-Old A/B switches were removed 2026-10-07, D532: GE_D204_OLD no
+    # (the -AB/-Old A/B switches were removed 2026-10-05, D532: GE_D204_OLD no
     # longer exists in the game, so they were documented no-ops)
 
     # Plumbing

tools_pc/capture_p7.sh

@@ -54,7 +54,11 @@ cleanup() {
   if [ "$EEP_HAD" = 1 ]; then cp "$EEP_ORIG" "$EEP" 2>/dev/null; else rm -f "$EEP" 2>/dev/null; fi
   rm -f "$INI_ORIG" "$EEP_ORIG" 2>/dev/null
 }
-trap cleanup EXIT INT TERM
+# D534: restore on EXIT only; INT/TERM just exit (which fires EXIT), so an
+# interrupted capture cannot continue past the restore with the snapshot gone.
+trap cleanup EXIT
+trap 'exit 130' INT
+trap 'exit 143' TERM
 
 # recipe comes from the tracked gate tool, not from this file
 eval "$(grep -m1 '^GOLDEN_DUMP=' "$VSH")"

tools_pc/golden/README.md

@@ -24,13 +24,13 @@ and per platform (`win`, and `linux` — 21 levels x 3 frames on each, captured
   D523, **superseded by D529**): the frame depends on the save's **content**, not
   its presence — D523's "presence, not which save" claim (worst_cell 0.175625 on
   each of the four candidate saves, 21.557% with none) was only ever measured on
-  Archives, and the 2026-10-07 A/B refutes it (facility 177.47 scene-level on the
+  Archives, and the 2026-10-05 A/B refutes it (facility 177.47 scene-level on the
   2026-10-05 playtest save, 0.0-0.25 on every other candidate on disk). `verify.sh`
   and `capture_p7.sh` install the canonical eep the same way they pin the ini
   (snapshot the local file if any, install the canonical one, restore-or-remove on
   exit), so the gate is reproducible on a fresh clone and in CI. A local playtest
   save is **never** the gate's input. The linux 63-frame set was re-captured
-  on the box under the canonical eep 2026-10-07 (D531; the 2026-10-05
+  on the box under the canonical eep 2026-10-05 (D531; the 2026-10-05
   re-round, D530, turned out to have run under the box-LOCAL save — the
   `capture_p7.sh` per-level re-pin bug, fixed in D531) and is valid.
 - Every run ends via `GE_QUITFRAME` (orderly quit; never a hard kill, D344).
@@ -66,7 +66,7 @@ capture with the recipe above and replace `<level>/<platform>/*.png`.
 
 ## History
 
-- **D531 (2026-10-07):** `capture_p7.sh`'s D529 adaptation was incomplete —
+- **D531 (2026-10-05):** `capture_p7.sh`'s D529 adaptation was incomplete —
   the per-level re-pin restored the pre-capture LOCAL save (D524-era line),
   silently overwriting D529's pre-loop canonical install, so the 2026-10-05
   box re-round (D530) ran under the box-local save, not the canonical one.
@@ -74,7 +74,7 @@ capture with the recipe above and replace `<level>/<platform>/*.png`.
   and re-captured: fresh 21/21 box round, 63 frames committed, replacing the
   D530 set (47/63 frames differ — save content demonstrably moves the frames;
   16 byte-identical save-independent scenes).
-- **D530 (2026-10-07, AMENDED BY D531):** the linux set was re-rounded on the
+- **D530 (2026-10-05, AMENDED BY D531):** the linux set was re-rounded on the
   box (X220, tree `442eb2ff`; `~/p7-reround.log` hit `P7 re-round DONE`
   2026-10-05 12:53 EDT, 21/21 rc=0) and committed, replacing the D525
   cleared-eep set (51/63 frames differ; 12 save-independent scenes byte-

tools_pc/verify.sh

@@ -112,7 +112,7 @@ while [ $# -gt 0 ]; do
     --script)   SCRIPT="$2"; USER_SCRIPT=1; shift 2 ;;
     --against)  AGAINST="$2"; shift 2 ;;
     --json)     JSON=1; shift ;;
-    -h|--help)  sed -n '2,32p' "$0"; exit 0 ;;
+    -h|--help)  sed -n '2,34p' "$0"; exit 0 ;;
     *)          POSARGS+=("$1"); shift ;;
   esac
 done
@@ -121,12 +121,12 @@ set -- "${POSARGS[@]:-}"
 case "${1:-}" in
   sweep)  MODE=sweep; shift; SWEEP_SUBSET="${*:-}" ;;
   parity) MODE=parity; LEVEL_ARG="${2:-}"; shift 2 2>/dev/null || true ;;
-  -h)     sed -n '2,32p' "$0"; exit 0 ;;
+  -h)     sed -n '2,34p' "$0"; exit 0 ;;
   "")
     # D529 never-do 4: a no-arg run must NEVER look like a green verdict. It
-    # prints the usage and exits NON-zero (2026-10-07: a no-arg pair was
+    # prints the usage and exits NON-zero (2026-10-05: a no-arg pair was
     # cited as a "double-green" 21/21 because it exited 0).
-    sed -n '2,32p' "$0"
+    sed -n '2,34p' "$0"
     echo "error: no mode given -- a sweep is 'verify.sh sweep', one level is 'verify.sh <level>' (never cite a no-arg run as a gate result; D529)" >&2
     exit 1 ;;
   *)      MODE=single; LEVEL_ARG="$1" ;;
@@ -257,14 +257,21 @@ pin_ini_640x480() {
 restore_ini() {
   if [ -n "$INI_BAK" ]; then cp "$INI_BAK" "$INI" && rm -f "$INI_BAK"
   elif [ "$INI_CREATED" = 1 ]; then rm -f "$INI"; fi
+  INI_BAK=""; INI_CREATED=0
 }
-trap 'restore_ini; restore_eep; kill_ours' EXIT INT TERM
+# D534: restore on EXIT only; INT/TERM just exit (which fires EXIT). A combined
+# `EXIT INT TERM` trap restored on Ctrl-C and then let the sweep CONTINUE --
+# the next level's pin re-installed the canonical save with the snapshot
+# already deleted, so the user's own ge007.eep was left replaced on disk.
+trap 'restore_ini; restore_eep; kill_ours' EXIT
+trap 'exit 130' INT
+trap 'exit 143' TERM
 
 # --- save pin (D523, SUPERSEDED by D529): the goldens were captured with a
-#     save file PRESENT. D529 (2026-10-07 A/B) proved the frame depends on the
+#     save file PRESENT. D529 (2026-10-05 A/B) proved the frame depends on the
 #     save's CONTENT, not just its presence -- D523's "presence, not which
 #     save" claim had only been measured on Archives (worst_cell 0.175625 on
-#     each of the four candidate saves, 21.557% with none), and the 2026-10-07
+#     each of the four candidate saves, 21.557% with none), and the 2026-10-05
 #     sweep (facility 177.47 scene-level on the local playtest save vs 0.0-0.25
 #     on every other candidate on disk) refutes it. The gate now INSTALLS THE
 #     CANONICAL SAVE (tools_pc/golden/ge007.eep, 2 KB, in-tree so a fresh
@@ -291,6 +298,7 @@ restore_eep() {
   [ -n "$EEP_PIN_BAK" ] || return 0
   if [ "$EEP_PIN_HAD" = 1 ]; then cp "$EEP_PIN_BAK" "$EEP_PIN"; else rm -f "$EEP_PIN"; fi
   rm -f "$EEP_PIN_BAK"
+  EEP_PIN_BAK=""; EEP_PIN_HAD=0
 }
 
 # --- run one level -> sets globals: R_STATUS R_FRAMES R_SYM R_BRIEF -------