Bucketing: Plot min/max bounds

[Chris Warburton]

We have plots for pseudorandom and recurrent-clustering performance; we
should also plot the min and max bounds.

These are currently fixed between sizes 1 to 10, since sizes 11+ are
currently infeasible to run. It *might* be possible to improve that, but
don't waste time trying to do it.

We should first plot these bounds on their own, so we can see what they
are.

We should then incorporate them into the plot of pseudorandom and
recurrent-clustering results.

We need to figure out the most appropriate way to represent this: maybe
just means?