Safer Redirects with Queue-Based URL Processing for Large Updates

Redirect handling has been updated to support safer, queue-based URL changes from operational flows. CSV product imports now detect actual changes in products_url (and related name-driven updates) and mark rows as pending redirects (redirect_status = 1) while copying the current URL into products_url_old. This now also covers the case where products_url is intentionally emptied, for example so AI translation can rebuild the slug later.

Important: if URL is emptied, redirect generation is deferred until a valid new URL exists, preventing broken redirects to empty targets.

Redirect creation is now processed in background cron batches (instead of direct/manual endpoint calls), with controlled throughput and completion marking after processing. This improves reliability for large updates and reduces immediate write pressure.

Category CSV updates were also aligned so relevant path/name changes can be flagged for redirect processing.

Recommendation remains: keep URLs stable; use redirects only when URL changes are truly needed.