Parallelise queries across "workers" by time range, for long queries

Now we have the query frontend, we should split queries than span multiple days into multiple queries and run them in parallel.

Special care will need to be take around handling vectors that span multiple days.