test(operad): regression coverage for the 4.0.4 member-of pair + port_color_map first use

[MSD21091969]

Closes #64. No behavior change — regression coverage only, pinning the enforcement that ffs0#172 (ontology 4.0.4) now depends on.

Why

member-of/has-member are absent from DefaultPortColors, so their color comes only from the port_color_map override — the first load-bearing use of that path. If the override merge, the WF02 pair loading, or the fail-closed color gate regresses, the 4.0.4 member-of LINKs would be silently rejected (or worse, silently weakened). Nothing tested that until now.

What (all four #64 asks, one new test file)

1. WF02 member-of pair — loader captures member-of/has-member (v4.0.4) alongside delegates-to (v3.13); ValidateLINK accepts both WF02 additional pairs and the primary governs, and rejects an undeclared variant with the declared set named.
2. port_color_map first use — the override merges onto DefaultPortColors (defaults survive), an ontology color wins over a built-in default, and an unknown color name or JSON null is a load error (a typo or None-emitting script must not alias onto the exempt ""). Empty string stays a documented exemption.
3. MOOS_INTEGRATION readback as a test — against the real sibling ontology, every port of every declared pair (primary + additional, all WFs) carries a color, and member-of is present + auth. The loader only warns on uncovered ports; this makes it a hard failure for the shipped ontology.
4. Interim manual check recorded in the file header (diff /operad/rewrite-categories + /operad/port-colors vs ontology.json at Doctor time) until (3) runs in CI.

Verification

• go test ./internal/operad/ — green (full hermetic suite)
• MOOS_INTEGRATION=1 go test ./internal/operad/ -run DeclaredPairsAllColored — PASS against ffs0 4.0.4
• go vet ./internal/operad/ clean · go build ./... ok

Reuses the shared writeOntology helper from loader_test.go (no duplicate).

🤖 Generated with Claude Code

[Copilot]

Pull request overview

Adds regression tests to the internal/operad ontology loader/validator surface to pin behavior required by the ffs0 ontology 4.0.4 upgrade (notably WF02’s member-of/has-member additional port pair and reliance on port_color_compatibility.port_color_map).

Changes:

• Adds a hermetic ontology fixture test ensuring WF02’s member-of additional pair is loaded into AdditionalPortPairs.
• Adds ValidateLINK coverage ensuring WF02 accepts member-of/has-member, delegates-to/delegated-by, and the primary governs pair, while rejecting an undeclared variant with a helpful error.
• Adds tests around port_color_map behavior (merge-over-defaults, override wins, unknown/null errors, empty-string exemption) plus a MOOS_INTEGRATION=1 check against the sibling real ontology.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[MSD21091969]

Agreed — removed the duplicated port_color_map unit tests. loader_test.go already covers override-wins / new-port / "" exemption / unknown / null (with synthetic ports). Kept only what was actually missing: the WF02 member-of + delegates-to pair (load + validate) and the real-ontology declared-vs-loaded readback that pins member-of's color specifically. File renamed to member_of_pair_test.go.