[supervisor] expose endpoints to collect workspace metrics

[akosyakov]

Since workspaces are ephemeral by nature we cannot poll them to collect metrics. Instead workspaces should increase counters on a new long living component which can expose them to prometheus. We could try to integrate https://github.com/prometheus/pushgateway into IDE proxy.

[iQQBot]

Instead of collecting the logs through ide-proxy, we decided to expose them directly to the Prometheus on the node through the supervisor