fix(local-stack): guard_port no longer kills make up on a fresh stack (#1076)

guard_port() assigns pid from `lsof ... | head -1` under set -euo
pipefail. lsof exits non-zero when nothing is listening on the port,
which is the normal case on the first `make up` of a fresh stack
(before the service has started). The non-zero status propagates out
of the command substitution and set -e kills the script before the
`[[ -z "$pid" ]] && return 0` emptiness guard runs, so the stack
never gets past the first guard_port call.

Append `|| true` so an empty result degrades to the guard instead of
aborting -- the same pattern the key/token lookups below already use.
The bug stays hidden once a service is already running, because
svc_running short-circuits before the lsof line.

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>

Author: yewreeka

dev/local-stack/stack.sh

@@ -88,7 +88,11 @@ svc_running() {
 guard_port() {
   local port="$1" name="$2" dir="$3" pid cmd
   svc_running "$name" "$dir" && return 0
-  pid="$(lsof -tnP -iTCP:"$port" -sTCP:LISTEN 2>/dev/null | head -1)"
+  # lsof exits non-zero when nothing is listening (the common case on a fresh
+  # stack before the service starts); under set -euo pipefail that would
+  # propagate out of the command substitution and kill the script before the
+  # emptiness guard. Same `|| true` pattern the key/token lookups below use.
+  pid="$(lsof -tnP -iTCP:"$port" -sTCP:LISTEN 2>/dev/null | head -1)" || true
   [[ -z "$pid" ]] && return 0
   cmd="$(ps -o command= -p "$pid" 2>/dev/null | head -c 140)"
   die "port ${port} is held by a foreign process (pid ${pid}: ${cmd}) — it would shadow ${name} and answer health checks with stale state. Kill it (kill ${pid}) and re-run."