Statistics and normal distibutions

[jkomoros]

• [x] Use for other random number distribution in other simulators
• [x] Add normal distribution
• [x] A helper like distributionOptionConfig() that returns an optionConfig that can be passed to linearDistribution(value, rnd) and do the distribution. In the future it could allow a distribution type in there, too. (But e.g. min/max should never be settable by a user but still pass with it...)
• [x] Support normal distribution (description, test that it works with multiple types in the generated config, and sample()) (maybe https://github.com/errcw/gaussian (it's not documented in readMe but you can pass a second parameter to random that is the rnd)
• [x] Add a 'fixed' value that just returns 'average'
• [ ] A power distribution *//Based on https://stats.stackexchange.com/a/406705
• [x] Simple mean, median, sum, of numbers for convenience
• [ ] Support a randomItem distribution with options
• [ ] Some way to have min, max, mean, median for all output metrics (instead of just enumerating them manually like we do now)
• [x] Support a round: 'ROUND', 'FLOOR', ''
• [x] Document how to use it
• [x] All other simulators use distribution
• [x] optionsConfig should get a hide(parentValues) -> bool function
• [x] It's annoying as an end user for a DistributionConfig, when you add it it's empty and you have to add a type AND the the proper types of arguments for that type. Ideally when you add it, it would have type=linear, average: average, already set (this would require a new state in the default/optional setup) (Now #51)
• [x] Add a min-max type
• [x] remove old linearDistributiuon

[jkomoros]

Make the distribution a global variable. Have a config that includes the type, the clipMin, clipMax, etc.

Have an optionsConfig and a withUserValues that extends for any missing values and then sample (rnd)

const performanceQualityDistribution = new LinearDistribution({name: 'performanceQuality', description:'The quality of the performance', step: 0.05, min: 0.0, max: 1.0});

//in optionsConfig:
{
  performanceQuality: performanceQualityDistribution.optionsConfig
}

And in the actual thing you need it, use the e.g.

performanceQualityDistribution.withUserValues(simOptions.performanceQuality).sample(rnd)

In the future, this would allow things like where you could have a Distribution that allows types to be picked by the user, where type is rendered as an option. e.g. LinearDistribution is just a Distribution, but with an option of types:['linear'], so the type selector is not rendered out in optionsConfig.

[jkomoros]

The one remaining minor todo isn't that important

[jkomoros]

Added a few TODOs so this s open again