Resolves #1935.

Adds a Quantiles aggregation that can be used to compute the quantiles of numeric data.

Notes

• Unfortunately, MongoDB has no builtin method for computing quantiles, so the implementation performs a full sort of the values. Thus quantiles() requires O(n) memory (DB-side, not client-side), not O(1) like other aggregations
• I benchmarked $group + JS sort versus $sort + $group on a dataset of 100K samples, and the JS sort version was faster
• I also tried https://stackoverflow.com/a/60694525/16823653, which was extremely slow

Example usage

import fiftyone as fo
import fiftyone.zoo as foz

dataset = foz.load_zoo_dataset("quickstart")

print(dataset.quantiles("uniqueness", [0.25, 0.50, 0.75]))
# [0.2202, 0.3377, 0.6255]

print(dataset.quantiles("uniqueness", 0.9))
# 0.6949

print(dataset.quantiles("predictions.detections.confidence", [0.25, 0.50, 0.75]))
# [0.0923, 0.2025, 0.5627]

print(dataset.quantiles("predictions.detections.confidence", 0.9))
# 0.9435