Benchmark nondeterminism carriers

[robrix]

We should benchmark the nondeterminism carriers (ideally including cull & cut).

I’d like to eventually construct our benchmarks to compare the space of programs using the various effects à la AutoBench, but until that happy day it’d be nice to at least cover something like n queens.

I think we should also benchmark state inside, outside, & on both sides of nondet, since that’s relevant to e.g. semantic’s program analysis.

[robrix]

Adding this to 0.6 because I want to benchmark #197 against the old church-encoded-list carrier.

[robrix]

Closed by #211.