Fixed: server error log no longer fills up when a marketplace returns a non-200
The scheduled check used the HTTP client's default behaviour, so any non-200 response from a resource page - a Cloudflare 403/429 served to your forum's server IP, a 5xx, and so on - was thrown and written to the server error log as a full stack trace on every cron run. (Thanks to the member who reported it.)
The check now treats a non-200 - and an unreachable or timed-out marketplace - as a normal "couldn't check this time" outcome: it's reflected in the coverage line and the last-checked time, without logging an exception. Genuinely unexpected errors are still logged.