Function dataflow::render::reduce::render_reduce_plan

fn render_reduce_plan<G, T>(
    plan: ReducePlan, 
    collection: Collection<G, (Row, Row)>, 
    err_input: Collection<G, DataflowError>, 
    key_arity: usize
) -> CollectionBundle<G, Row, T> where
    G: Scope,
    G::Timestamp: Lattice + Refines<T>,
    T: Timestamp + Lattice, 

Render a dataflow based on the provided plan.

The output will be an arrangements that looks the same as if we just had a single reduce operator computing everything together, and this arrangement can also be re-used.