feat(comid): enforce acyclic constraint on dependency-triples

[rsampaio]

Per draft-ietf-rats-corim 5.1.11.2, a domain-id MUST NOT appear in its trustees list. Add the check to DomainDependencyTriple.Valid() and extend the test suite accordingly.

[thomas-fossati]

Thanks very much! LGTM.

I have a couple of comments which could be addressed either in this PR or in a future changeset:

1. The DomainDependencyTriple::Valid function could also check for duplicate trustees
2. DomainDependencyTriples::Valid could do more than call the Valid function for individual triples; it could also check the overall consistency of the dependency graph expressed by all the DDTs present and flag cycles across triples.

[rsampaio]

@thomas-fossati good call on the second validation, I added here since it is open and relevant to recent changes! Thanks for the feedback!

[thomas-fossati]

Thank you again!

Your cycle-detection choice is an interesting one. Personally, I’d have used topo sorting and, because I am a lazy individual, I’d have pilfered one of the existing packages (e.g., stevenle/topsort/v2 or similar) :-)

[rsampaio]

Thank you again!

Your cycle-detection choice is an interesting one. Personally, I’d have used topo sorting and, because I am a lazy individual, I’d have pilfered one of the existing packages (e.g., stevenle/topsort/v2 or similar) :-)

I went with DFS because it was compact enough for this use case, agree that topo sorting would also work :)