Implement better retry logic on failed queries

[alexanderbez]

Currently the worker naively retries every failed export:

// start starts a worker by listening for new jobs (block heights) from the
// given worker queue. Any failed job is logged and re-enqueued.
func (w worker) start() {
	for i := range w.queue {
		if err := w.process(i); err != nil {
			// re-enqueue any failed job
			// TODO: Implement exponential backoff or max retries for a block height.
			go func() {
				log.Printf("re-enqueueing failed block %d\n", i)
				w.queue <- i
			}()
		}
	}
}

A better approach is needed to not continuously retry the same heights.