Add benchmark that makes measurable use of enum

[mdboom]

We had an unexpected regression in enum in the 3.11 cycle and we know from running the profiler over pyperformance that none of the benchmarks spend a significant amount of time dealing with enums.

We should try to add a good benchmark that makes meaningful use of enums, ideally not just an enum microbenchmark, though.

[kmod]

Shameless plug -- it might be less than you want, but the pyston djangocms and flaskblogging benchmarks spend about 0.5-1% of their time in enum.py

[mdboom]

That's actually really helpful, @kmod. I had been struggling to find something (just by poking around with Github code search). This might get resolved by finishing #175 then.