feat: add tdigest data structure for statistics

[AlSchlo]

This pull request implements a simplified version of Ted Dunning's TDigest algorithm for efficient quantile/cdf computations.

It supports a fully parallelizable, memory-bounded computation scheme, along with an easy API.

Simplified for two reasons:

• The linear interpolation is only done for the quantile, not the cdf.
• Linear interpolation in general could be done more efficiently if we leverage unit-weighted centroids at the edges, as described in the paper.

Both are marked as TODOs in the code for future work.

Most open-source implementations found online had bugs, were incomplete, or were too complex (i.e., poorly written).

Includes unit tests on uniform and weighted distributions.

Next steps: Implementing the HyperLogLog algorithm for NDistinct.

[yliang412]

I haven't read the paper in 100% detail, but overall LGTM!

Could we keep the license consistent with the one for the entire repository (LICENSE)? I also suggest that you could add a short description for the gungnir crate under the structure section in the repo README.

[yliang412]

You could use doc comments i.e. /// above the structs and fields so the comments show up in the documentation. For example,

/// The TDigest structure for the statistical aggregator to query quantiles.
pub struct TDigest {
    /// A sorted array of Centroids, according to their mean.
    centroids: Vec<Centroid>, 
...
}

[AlSchlo]

Where did I change the License? I don't think I added anything about the license.
Do you mean the cool header at the top of each Gungnir™ file? 😂

[yliang412]

Oh yea. I was talking about those. Those are cool though lol.