fix: use wait -n to reap zombie subshells in parallel job pool

kill -0 returns success for zombie processes (completed but not yet reaped).
Without wait -n, _PIDS never shrinks below PARALLEL_TASKS after all initial
slots fill — remaining tasks never get submitted.

wait -n on the PIDS list blocks until one child completes and reaps it,
allowing kill -0 to correctly detect the freed slot.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Author: sjarmak

configs/run_selected_tasks.sh

@@ -307,15 +307,19 @@ is_task_completed() {
 # ============================================
 declare -a _PIDS=()
 
-# Block until fewer than PARALLEL_TASKS jobs are running
+# Block until fewer than PARALLEL_TASKS jobs are running.
+# Uses wait -n to reap zombies (kill -0 sees zombies as alive; without reaping,
+# slots never free even after child processes complete).
 _wait_for_slot() {
     while [ "${#_PIDS[@]}" -ge "${PARALLEL_TASKS}" ]; do
+        # Reap any completed child (non-blocking: wait for exactly one child to finish)
+        wait -n "${_PIDS[@]}" 2>/dev/null || true
+        # Rebuild PID list, excluding any that are no longer alive
         local new_pids=() p
         for p in "${_PIDS[@]}"; do
             kill -0 "$p" 2>/dev/null && new_pids+=("$p")
         done
         _PIDS=("${new_pids[@]}")
-        [ "${#_PIDS[@]}" -ge "${PARALLEL_TASKS}" ] && sleep 2
     done
 }