Added two new tests with duplicate triples

[iherman]

This PR is to fix #191.

Adds two tests: each contain two identical triples in the input (without and with an identical blank node, respectively); the parsing of the the nq files should make it sure that a dataset is canonicalized, ie, the duplication of triples disappear.

[iherman]

@gkellogg I do not know the details of the github action that reports an error. I would leave that in your able hands...

[gkellogg]

Without objection, I think we should merge this and allow updated implementation reports to be submitted.

[davidlehn]

rdf-canonize passes the tests here and had a local test for this already. It used a variation of the tests here with subject and object blank nodes. Should that be added here too or is it redundant?

Input:

_:s <https://www.example.org/p> _:o .
_:s <https://www.example.org/p> _:o .

Output:

_:c14n1 <ex:p> _:c14n0 .

[gkellogg]

If those tests would add something not otherwise caught, then we should add them to this PR.

[iherman]

If those tests would add something not otherwise caught, then we should add them to this PR.

True. But I am not sure that tests really add something. The issue to be tested is whether the implementation correctly filters duplicate quads in an nquads representation of a graph, whether those quads/triples include (identical) blank nodes or not. The fact that there is one bnode or there are two does not change the situation imho.

[yamdan]

My implementation also passes the added two tests. I will update the report after the PR is merged.

[pchampin]

FTR, my implementation also passes these tests