feat: New schema for HS3 Support

[kratsg]

Runtime validations:

• check that staterror is only defined once per sample
• check that if staterror is defined, that the sample has defined errors
• check that contents and errors have the same shape if errors is defined
• check that likelihoods.distributions has same shape as likelihoods.datasets
• check that measurements.likelihood points to a valid likelihood
• check that measurements.domains points to a valid domain
• check that measurements.inits points to a valid init
• made modifier data optional (in cases where it was null)

Accepted Defaults:

• axes has min=0, max=1 (or edges = [0, 1])
• parameter/parameters is just the name of the parameter(s)
• staterror data is moved to the sample errors

To Dos:

• drop luminosity when customizable constraints can be supported (feat: Configurable constraints per-modifier #1829)
• support schema migration (feat: HiFa v1.1.0 schema development #1978)
• fix Multiple staterror modifiers per sample and channel, round trips and duplicate modifiers #1830
• determine parameter naming for a given systematic (to fix inits and domains)
• determine the correct naming for measurements key, and for datasets key
• clarify the usage of axis in HS3 (binned vs unbinned)
• clarify measurement.poi vs measurement.parameters_of_interest
• modifier.parameters to modifier.parameter (for pyhf) [since we don't have per-parameter naming]
• move version into metadata.version

References:

• differences to pyhf JSON hep-statistics-serialization-standard/hep-statistics-serialization-standard#30
• https://github.com/root-project/root/blob/master/tutorials/roofit/rf515_hfJSON.json

Pull Request Description

Please first read CONTRIBUTING.md.

Please describe the purpose of this pull request in some detail and what the specific feature being added will do. Reference and link to any relevant issues or pull requests (such as the issue in which this feature was first suggested).

Checklist Before Requesting Reviewer

• Tests are passing
• "WIP" removed from the title of the pull request
• Selected an Assignee for the PR to be responsible for the log summary

Before Merging

For the PR Assignees:

• Summarize commit messages into a comprehensive review of the PR

[alexander-held]

Is this the object that goes into parameter_points in the current HS3 draft? I don't quite follow how to map this to the document. I started looking at this as I as wondering if the format would allow to conveniently distinguish between initial parameter values for the purpose of minimization starting point and default values for the purpose of dataset generation.