Changing precision?

[mnutt]

We've been using node-hash-ring very successfully for many years, so thank you for writing this library.

One thing we noticed was that the distributions, while even enough for balancing servers, often aren't even enough for statistical purposes. The use case is using it to distribute millions of user identifiers into a few buckets, and when doing so we sometimes see up to 5% differences in group sizes.

Would you be interested in a PR for an optional argument for tuning the number of rings, to trade off some cpu/memory for more even distribution?