Don't die just because some posts failed to render

For god knows which time already pleroma starts timing out while FE tries to fetch timeline. This leaves FE completely broken and unusable until admin fixes the problem and restarts the whole thing.

Some suggestions:

1. Have an internal timeout for rendering a request and each post.
2. When request timeout is upon us, just render what's ready, drop every unrendered post (put them into queue to re-render in safe mode* maybe?)
   alternatively: track if request timed out and re-render entire request in safe mode next time?
3. Log the long-running post renders/request as errors so it's easier to debug them.

*by safe mode i mean render post ignoring stuff that might fail like rich media previews.