Refactor/cleanup the archive reader/writer code

What's the issue?
The archive implementation has grown organically over time and has become complex in some places and "bodged" in others.

For example, we've got functions at the archive level which access/fetch information from specific backups; ideally, we've have some sort of hierarchical reader/writer logic where each level handles logic required at each level.

This work will massively simplify things when the time comes to support workload distribution/fine grained locking.

We also have a lot of functions which are complex/difficult to follow we should break these down into smaller logical components where possible.