Update cache in background
Because of the synchronous nature of the API we run in situations where the limited availability of web workers causes the federation to fail. This seems to happen more likely when the cache information about remote fediverse actors is not hot anymore.
A user on Host A tries to playback a song from Host B. Host A asks Host B for the track. In order to verify the requests, Host B needs up-to-date information about the fediverse actor. If its cached, its no problem at all. If this cache is outdated, we get some trouble since Host B asks Host A for the profile information. At least one web runner on Host A is busy with requesting the track from Host B. So its more likely the update request from Host B will timeout.
We fixed this for #1303 (closed) by suggesting to increase the number of web workers. Anyway, I think we should increase stability of federation by refetching these actor profile data in the background using the celeryworker.