This saves us one container network in the ideal case.

`matrix_homeserver_container_url` is potentially wrong in certain
scenarios (going through `matrix-homeserver-proxy`).

This is a bit of a compatibility break.
The role was defaulting the Postgres password to `some-password` and we
auto-generate it now.

However, rebuilding both Postgres and this service should unify the
database credentials and the service configs to the new value.

This service will be removed soon, but for now we need to get rid of
`matrix_docker_network` usage everywhere.

Same as 250b91a40968e, but for Slack

Postgres is not in `matrix_docker_network` anymore, so what we had
before could not possibly work anymore.

The old variables existed as well, but I inteded to use these new ones.

The user may be running Prometheus elsewhere. It doesn't need to be
getting installed using the current playbook.

Regression since ce2f541deb