Copilot task: normalize GraphLab functional lab manifest, validation, and release-dry-run skeleton

[mdheller]

Mission

Normalize graphlab as a SociOS functional ML lab surface for graph embeddings, ontology alignment, semantic graph extraction, graph neural experiments, and knowledge-graph model tuning.

This issue is designed for a Copilot coding agent. Keep the change bounded and open a PR.

Scope

Add a minimal lab skeleton without pulling heavy model dependencies:

• docs/LAB_SURFACE.md
• docs/PROPHET_INTEGRATION.md
• examples/functional-service.json
• examples/eval-fixture.json
• tools/validate_examples.py
• Makefile
• optional .github/workflows/validate.yml

Required Make targets

make validate
make emit-service-manifest
make emit-eval-fixture
make release-dry-run

Required semantics

• GraphLab is a lab surface, not the platform runtime service.
• Platform service promotion belongs to SocioProphet/prophet-platform and governance ledgers.
• SourceOS carries signed service refs only; it must not carry mutable model authority.
• Examples must be deterministic and small.

Acceptance criteria

• make validate passes with no network or model downloads.
• make emit-service-manifest writes or prints the functional service manifest.
• make emit-eval-fixture writes or prints the eval fixture.
• Docs explain integration with functional-model-surfaces, prophet-platform, sourceos-model-carry, prophet-core-query, ontogenesis, and semantic-serdes.

Do not

• Do not download or vendor model artifacts.
• Do not add GPU-heavy dependencies.
• Do not implement real training jobs.
• Do not modify unrelated files.