Implement `config`, `sysinfo` and `clusters` verbs

[lars-t-hansen]

See #604 (comment). For the time being we just need something basic. Even clusters is optional, really, but it's clearly a missing piece of the infrastructure.

Thinking about this some more, I'm seeing the contours of something general (related to #517 actually):

• there is a clusters table and sonalyze clusters queries this
• there is a sysinfo table and sonalyze nodes queries this (can't use sysinfo as the verb at the moment b/c already taken for old sysinfo ingest point, we can fix this over time)
• there is config table and sonalyze config queries this
• there is an sacct table and sonalyze sacct queries this (already implemented)
• there is a sample table and sonalyze parse queries this (probably add sample as alias) (already implemented)

Then there are jobs, load, uptime, top, profile, metadata. These are built-in aggregators that currently work only on Sonar + config data but could incorporate Slurm data and sysinfo data - something to consider over time. But for the time being we just leave them alone and maybe document them slightly better.