Define cross-language evaluation semantics

[uriahf]

Goal

Establish the Stage 0 semantic vocabulary for the cross-language parity work before adding characterization tests.

Contract

Defines shared meanings for:

• model
• population
• evaluation (model × population)
• evaluation context
• fixed time horizon
• plotted series
• global, population-specific, and population-and-horizon-specific reference-line scope

It also records the key compatibility rule that existing reference_group terminology is a generic implementation/grouping concept and is not the target domain abstraction. Stage 1 must characterize what it represents in each current code path rather than assuming it means either model or population.

Scope

Documentation only. No statistical calculations, public APIs, input formats, return values, plotting behavior, labels, colors, reference lines, package metadata, generated documentation, or rtichoke_viz are changed.

Paired PR

Python: uriahf/rtichoke_python#364

Both PRs carry the same semantic contract so Stage 1 tests can use one vocabulary in both repositories.