Merge connected worker stats with job stats in a single persisted collection

[rhyslbw]

Now

• The job server stats are stored in an in-memory collection as it's then unlike to ever be out of sync when the process restarts, and naturally supports multiple instances.
• The connected workers are tracked in a separate collection and offered in a separate publication. This was done since all connected workers across all app instances must be collected.

Proposal

• Make JobServer stats persistent since it's also relevant to a single user in the front-end how many total jobServer instances are online, and how many workers are connected to each.
• Need to use a persistent instance identifier to ensure a process exit or reconnection event is handled and no zombie records are left. Most likely this will be an ENV var