//! Crawling: the [`spider::Spider`] trait, the [`engine::CrawlerEngine`]
//! that runs it, and the supporting machinery — scheduling with request
//! dedup, robots.txt compliance, AutoThrottle, checkpoints for
//! pause/resume, a development-mode response cache, link extraction, and
//! crawl results/exports.