check: add ./check stress stage; fix unreachable bench/mount dispatch

- ./check stress spins a fresh cube-server+cubec built from this tree on a
  throwaway socket/store and drives ~150s of real prog/run traffic while
  sampling the per-command latency + per-C telemetry.
- tools/stress.sh: never touches the production daemon; STRESS_SECONDS override.
- Fixed latent control-flow bug: the old '|| exit 0' guards made 'bench' and
  'mount' subcommands unreachable after the gate (they exited early). Now uses
  explicit run_* flags and only exits at the true end.

tools/stress.sh

@@ -0,0 +1,64 @@
+#!/usr/bin/env bash
+# Sustained stress stage for CUBELinux-2 (invoked by `./check stress`).
+#
+# Builds cube-server + cubec from THIS workspace tree, spins a fresh daemon on a
+# throwaway socket + store, then drives it hard for STRESS_SECONDS (default 150)
+# with real prog/run traffic while sampling the per-command latency + per-C
+# namespace telemetry. This exercises the new cubesys `stats` path under load.
+#
+# It deliberately does NOT touch the production daemon (real socket/store in
+# /home/luulu/.cubelinux) so it is safe to run any time. Numbers are
+# informational; the stage passes iff the gate passed and the daemon survives.
+set -uo pipefail
+REPO="$(cd "$(dirname "$0")/.." && pwd)"
+cd "$REPO"
+export PATH="$HOME/.cargo/bin:/usr/local/bin:/usr/bin:/bin:$PATH"
+command -v cargo >/dev/null || { echo "cargo not found" >&2; exit 127; }
+
+JOBS="${JOBS:-4}"
+SECS="${STRESS_SECONDS:-150}"
+
+echo "building cube-server + cubec from tree ($REPO) ..."
+cargo build -p cubesys --release --bins --jobs "$JOBS" >/dev/null 2>&1 || {
+  echo "BUILD FAILED (cubesys bins)"; exit 1
+}
+SRV="$REPO/target/release/cube-server"
+CLI="$REPO/target/release/cubec"
+[ -x "$SRV" ] && [ -x "$CLI" ] || { echo "binaries missing after build"; exit 1; }
+
+TMPD="$(mktemp -d /tmp/cube-stress.XXXXXX)"
+SOCK="$TMPD/cube.sock"
+STORE="$TMPD/store.json"
+trap 'kill $DPID 2>/dev/null; wait $DPID 2>/dev/null; rm -rf "$TMPD"' EXIT
+
+echo "starting stress daemon on $SOCK ..."
+"$SRV" --socket "$SOCK" --store "$STORE" >"$TMPD/daemon.log" 2>&1 &
+DPID=$!
+# wait for socket (max ~15s)
+for i in $(seq 1 15); do [ -S "$SOCK" ] && break; sleep 1; done
+[ -S "$SOCK" ] || { echo "daemon failed to start (see $TMPD/daemon.log)"; kill $DPID 2>/dev/null; exit 1; }
+
+START=$(date +%s)
+END=$((START + SECS))
+PAIRS=0
+SAMPLE_EVERY=25          # sample stats every 25 pairs (~ once a second at load)
+while [ "$(date +%s)" -lt "$END" ]; do
+  X=$((PAIRS % 256))
+  "$CLI" --socket "$SOCK" "prog /c077/z001/y001/x$X const $((PAIRS % 97)) halt" >/dev/null 2>&1
+  "$CLI" --socket "$SOCK" "run  /c077/z001/y001/x$X"                    >/dev/null 2>&1
+  PAIRS=$((PAIRS + 1))
+  if [ "$((PAIRS % SAMPLE_EVERY))" -eq 0 ]; then
+    echo "--- [$(date -u +%H:%M:%S)] stress sample $((PAIRS / SAMPLE_EVERY)) (pairs=$PAIRS) ---"
+    "$CLI" --socket "$SOCK" "stats" || true
+  fi
+done
+
+# final health snapshot
+echo "--- final stats (pairs driven: $PAIRS over ~${SECS}s) ---"
+"$CLI" --socket "$SOCK" "stats" || true
+
+ELAPSED=$(($(date +%s) - START))
+if [ "$ELAPSED" -lt 1 ]; then ELAPSED=1; fi
+echo "stress throughput: ~$((PAIRS / ELAPSED)) prog+run pairs/s (informational)"
+echo "STRESS STAGE OK (daemon alive, $PAIRS pairs driven)"
+exit 0