Factor aggregation job handling to seperate state machine

[mendess]

By having the logic to handle an aggregation job live in a state machine like this instead of having just a top level function that does it all, it's easier to give users of the library ways to customize where and when the y want to execute each part necessary in the aggregation of report shares, without compromising on the correctness of these.

This PR only transforms helper::handle_agg_job_req into a state machine, but the other top level functions could also be transformed in similar ways. Future PRs will use this to eventually implement the async handling of aggregation jobs