Implement `t-digest` algorithm for online histogram calculation

[tasansal]

This would be used to calculate dataset histogram during ingestion. t-digest algorithm is very popular (especially in map-reduce ops in Apache Spark).

Computing Extremely Accurate Quantiles Using t-Digests (Dunning & Ertl 2019):
https://arxiv.org/abs/1902.04023

Some light weight explainer:
https://www.gresearch.com/news/approximate-percentiles-with-t-digests

There are two python libraries that do it:

• tdigest: https://github.com/tdunning/t-digest
• tdigest-rs (🦀): https://github.com/G-Research/tdigest-rs

E.g. each distributed worker calculates their part of the data distribution representation. At the end, they would be combined for the final histogram.