Define the Workshop conformance result contract and stable feature identities

[Teakowa]

Parent: #10

Goal

Define the canonical machine-readable conformance result model and stable Workshop feature identities that later census, regression, seasonal, and downstream-provider evidence can share.

Context

workshop-rs now provides the released canonical Workshop semantic core, but current tests are primarily repository-local focused gates. #10 requires failures and compatibility evidence to be attributable to concrete Workshop capabilities rather than only files, snapshots, or whole-script pass/fail results.

This issue establishes the contract first so later corpus work does not invent incompatible schemas independently.

Scope

• Define stable, locale-independent feature identities for the currently declared Workshop surface, including events, actions, values, enums/content IDs, settings, variables, subroutines, control flow, strings/localization, and other catalog/WIR-backed constructs where evidence exists.
• Define a machine-readable conformance result model that distinguishes at least matched behavior, unsupported surface, known gap, unexpected regression/divergence, and inconclusive evidence.
• Define evidence/provenance fields sufficient to identify the fixture/source, evidence class, catalog identity, implementation version/revision where relevant, locale, and external/live-client evidence where applicable.
• Keep feature identity separate from localized spelling and from source-language provider identities.
• Define semantic/normalized equivalence as the default comparison contract where textual identity is not observable semantics.
• Add focused tests for schema stability, serialization, invalid states, and representative feature identities.
• Document the public/reusable parts of the contract that opy-rs and del-rs may consume without transferring Workshop ownership.

Non-goals

• Building the complete feature census fixture.
• Importing real projects or live-client captures.
• Designing OPY/DEL-specific compatibility result schemas.
• Freezing a general cross-repository testing framework or plugin ABI.
• Treating every catalog entry as a separate implementation issue.

Acceptance criteria

• Conformance results can report the required states without counting known gaps or unsupported cases as passes.
• Representative Workshop capabilities have stable machine-readable feature IDs independent of locale and provider naming.
• Evidence/provenance metadata can distinguish synthetic, regression, real-project, and live-client evidence without relying on implementation-generated expectations as the oracle.
• The contract has serialization/validation tests and concise documentation.
• Later Establish a tests-first Workshop corpus and seasonal feature census #10 work can build census and corpus evidence on this contract without redefining the identity/result model.