Improve statistics interface

[jmid]

To inspect the distribution of generators QCheck offers

• Gen.generate{,1} to observe samples
• set_collect to label each sample with a string
• add_stat and set_stats to print a histogram and calculate avg/stddev/... based on some numeric property (e.g., List.length)

As part of reworking the generator and extending our existing tests of Stdlib.Sys I found myself

• reading about the Elm PBT port's abilities in this direction: https://martin.janiczek.cz/2025/05/01/elm-test-distributions.html
• watching “Building on developers’ intuitions to create effective property-based tests” https://www.youtube.com/watch?v=NcJOiQlzlXQ

Based on the above, I found myself

• wanting a generalization of set_collect that would allow to describe and summarize several separately specified labelings for one test - currently set_collect is singular, whereas set_stats is plural (and add_stat lets us add more)

and envious about

• labelExamples - to observe a sample from a label (e.g., "What's an example classified as a non-empty list?")
• checkCoverage - to ensure a certain distribution (e.g., "At lease 10% of the test input is Some with a non-empty list")

along the lines of the Haskell and Elm APIs:
https://hackage.haskell.org/package/QuickCheck-2.15.0.1/docs/Test-QuickCheck.html#g:21
https://package.elm-lang.org/packages/elm-explorations/test/2.2.0/Fuzz#labelExamples
https://package.elm-lang.org/packages/elm-explorations/test/2.2.0/Test-Distribution
https://package.elm-lang.org/packages/elm-explorations/test/2.2.0/Test#reportDistribution