Automated replica setup and recovery

Nobody has claimed this yet.

Currently, there's no automated way to set up a new replica or to replace a broken replica. A user has to perform `rsync` from an existing healthy replica's data directory using command-line interface, which is obviously error-prone.

We should implement internal remote API:

- restricted to inter-replica communication
- dedicated to replica setup and recovery

.. so that configuring a new replica is as simple as running a single command. For example:

```
$ cd centraldogma/bin
$ ./setup-replica 192.168.0.100:36462 # the IP address and port of an existing healthy replica
... retrieves conf/* and lib/ext/* ...
... generates UUID and writes to conf/dogma.json ...
... retrieves all repository data ...
$ ./startup
... a new replica joins the cluster ...
```

In case of recovery, you could simply wipe out the existing `data` and `conf` directory and run `setup-replica`.

Use centraldogma/bin/setup-replica and startup as the named entry points; trace the existing inter-replica communication and the conf, lib/ext, and data layouts before deciding where setup and recovery belong. Done means a single setup-replica command can retrieve the required configuration and repository data, generate a UUID in conf/dogma.json, and let the new replica join after startup.

Written by the indexing model from the issue text.