Deduplicate GNM edge list

[RobbieRao]

Summary

• derive deduplication keys from canonicalized undirected edge endpoints
• remove duplicate directed edges from GNM.edge_list
• add a regression assertion that every returned directed edge is unique

Root cause

edge_list canonicalized each edge as (min(v1, v2), max(v1, v2)), but computed uniqueness keys from the original directed endpoints. Shared quad edges commonly appear with opposite winding, so (u, v) and (v, u) received different keys even though they represented the same undirected edge. Both copies were then expanded in both directions.

Impact

For GNM Head v3, edge_list shrinks from 141,296 rows to 70,946 rows while preserving the same adjacency. All 70,946 returned directed edges are unique.

Testing

• python gnm/shape/gnm_numpy_test.py — 82 passed, 2 skipped
• python gnm/shape/run_all_tests.py — 278 passed, 2 skipped
• pylint gnm/shape/gnm_xnp.py gnm/shape/gnm_numpy_test.py

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.

[RobbieRao]

@googlebot rescan