Split out the workers
---

The current vm resource is
```bash
flyctl machine show
```
Inside the one machine for `app` process group, there's supervisord, then foreman, managing an unicorn `web.1` (master + 2workers), and the rufus scheduler `jobs.1` task. the unicorn is only useful when i open the site, or inoreader comes to fetch the feed. rufus should be running all the time, to execute scheduled tasks (agents, events propagation, and cleanups). Let's split out `app` to `web` and `jobs`, then we can let web autostop/start according to load.

 # Virtual1

# Virtual1
# eDP-1

workspace 5 output $side $main

workspace 10 output $side $main

workspace 3 output $side $main
assign [class="(?i)emacs"] 3
assign [app_id="(?i)emacs"] 3

workspace 3 output $main
assign [class="(?i)emacs"] 3
assign [app_id="(?i)emacs"] 3

workspace 9 output $side $main

workspace 9 output $main