Support custom loop factory in bench_async_func

[itamaro]

It may be desirable to use a custom event loop in an async benchmark (e.g. for benchmarking python/cpython#97696)

bench_async_func currently doesn't support that - it will always use the default loop