Allow Googlebot to access `/api/graphql`
While crawling GitLab.com, the Googlebot loads each page including dynamic resources from /api/graphql.
At present though, it will not access this endpoint as robots.txt is configured to block Googlebot.
This can lead to some of the rendering problems documented in #21303 (closed).
We should allow the Googlebot to access /api/graphql while blocking all other /api/ endpoints from webcrawlers.
This should be safe to do, since the endpoint is a POST and crawlers would not (afaik) access POST resources in any manner other than from the javascript on a page.
Related slack thread (while it lasts): https://gitlab.slack.com/archives/CPAPAKKA7/p1580196600002200