Automatically fetch runs from archives

[as2388]

Automatically fetches runs from archives when necessary. Works as a drop-in replacement for previous versions of Rapid Pro Tools in the pipelines.

Main caveat is that it does some unnecessary work, e.g.:

• Will fetch all archives once for each flow, so expensive when starting up.
• Fetches archives known to be empty.
• Pipelines request all runs since the modification date of the most recent run in that flow when running incrementally. This means for the oldest runs, time will be wasted searching in more recent archives, because the most recent update will fall in the archive period, which will hurt the performance of incremental fetching.

But better to get something that works than pre-optimise.

Takes <1 hour to fetch all data for WorldBank (7 flows from 2 different TextIt instances).

[IsaackMwenda]

Thanks @as2388 .