We need to fix how timeouts are set in mediawiki-config too.

The timeout in the config is set in set-time-limit and is based on the host header. When we switch to generic MediaWiki apache configuration for job runners, we would need to use the Host header for selecting the specific wiki, just like normal app servers do. Thus, the code referenced above would no longer work.

Apache sets SERVERGROUP variable we could use to differentiate job runners from other app servers, but we can't use that one either, because both video scaling and general job running is happening on the same set of hosts in reality, thus SERVERGROUP will be the same for both, however the timeout needs to be different.

We could add support for Request-Timeout HTTP header and set it as a timeout, but for job runners only. We probably don't want client requests to be able to override the request timeouts.