Implement Bag of Little Bootstraps

[mdekstrand]

To support statistical analysis (#784) and distributional analysis (#414), it would be useful to support the Bag of Little Bootstraps. Since there does not seem to be an implementation in any of the standard Python statistics packages, we should implement one under lenskit.stats.

The first implementation should be sequential. A parallel implementation with Ray (or with free-threaded Python) would be a good idea, but we can do sequential first.