It is not meaningless according to benchmarks.

ConsumerWorkService is meant to guarantee ordered dispatch of delivered protocol methods on the given channel. The decision to pick MAX_RUNNABLE_BLOCK_SIZE goes back to 2010. If you have
specific suggestions as to what can be improved and can prove the effectiveness with benchmarks,
we would happily accept a PR.

The value can be set to 128 or 256, the effectiveness of higher prefetch values hits the law of diminishing returns.

Setting the value to 256 produces a low single digit % gain in throughput and a comparable reduction in mean consumer latency. I will submit a PR in a bit.