set up loki to log on production and staging

I vaguely remember that it was set up in `/etc/docker/daemon.json` and looks like
```
{
"default-runtime": "nvidia",
"log-driver": "loki",
"log-opts": {
"loki-url": "http://[some IP address for loki]:3100/loki/api/v1/push"
},
"runtimes": {
"nvidia": {
"args": [],
"path": "/usr/bin/nvidia-container-runtime"
}
}
}

```
but we had trouble running Rodan on new instances, so we removed all settings that do not seem to affect main functions, including kubernetes and loki. We need to put loki back at least to collect useful logs.