feat: make pendings job processing resilient to connection loss

Instead of passing a single connection into the pending job processor,
use the connection pool to pull a new connection on every loop, releasing
it after every use.

backends/postgres/postgres_backend.go

@@ -191,11 +191,7 @@ func Backend(ctx context.Context, opts ...neoq.ConfigOption) (pb neoq.Neoq, err
 	go p.listenerManager(ctx)
 
 	// monitor queues for pending jobs, so neoq is resilient to LISTEN disconnects and reconnections
-	pendingJobsConn, err := p.pool.Acquire(ctx)
-	if err != nil {
-		return nil, fmt.Errorf("unable to get a database connection: %w", err)
-	}
-	p.processPendingJobs(ctx, pendingJobsConn)
+	p.processPendingJobs(ctx)
 
 	p.listenConnDown <- true
 
@@ -831,18 +827,27 @@ func (p *PgBackend) announceJob(ctx context.Context, queue, jobID string) {
 //
 // Past due jobs are fetched on the interval [neoq.DefaultPendingJobFetchInterval]
 // nolint: cyclop
-func (p *PgBackend) processPendingJobs(ctx context.Context, conn *pgxpool.Conn) {
+func (p *PgBackend) processPendingJobs(ctx context.Context) {
 	go func(ctx context.Context) {
-		defer conn.Release()
+		var err error
+		var conn *pgxpool.Conn
+		var pendingJobs []*jobs.Job
 		ticker := time.NewTicker(neoq.DefaultPendingJobFetchInterval)
 
 		// check for pending jobs on an interval until the context is canceled
 		for {
-			pendingJobs, err := p.getPendingJobs(ctx, conn)
+			conn, err = p.acquire(ctx)
+			if err != nil {
+				p.logger.Error("[pending_jobs] unable to get database connection", slog.Any("error", err))
+				<-ticker.C
+				continue
+			}
+
+			pendingJobs, err = p.getPendingJobs(ctx, conn)
+			conn.Release()
 			if errors.Is(err, context.Canceled) {
 				return
 			}
-
 			if err != nil && !errors.Is(err, pgx.ErrNoRows) {
 				p.logger.Error(
 					"failed to fetch pending jobs",