test(discovery): fix bootstrap-selector test left red by #941

[seonghobae]

Why

main currently has one failing test. #941 ("fix(discovery): keep credential catalogs independent", merged 42da1d5) removed _provider_family() — the heuristic that collapsed nvidia_nim/nvidia_nim_sub into one "outage domain" — across model_discovery.py, provider_bootstrap.py, and provider_catalog_bootstrap.py, correctly updating the tests in those three areas. It didn't touch tests/test_discovery_bootstrap_selection.py, which has its own separate test asserting the old collapsing behavior via select_bootstrap_discovered_agents:

$ python -m pytest tests/test_discovery_bootstrap_selection.py -q
FAILED tests/test_discovery_bootstrap_selection.py::test_bootstrap_selector_treats_nim_primary_and_sub_as_one_outage_domain

Confirmed this is red on unchanged origin/main (not introduced by any other branch).

What changed

Updated the one test to assert the now-correct behavior: nvidia_nim and nvidia_nim_sub are independent KV credential boundaries, so cheapest-first bootstrap selection admits both before a pricier, unrelated provider — matching the exact pattern #941 already applied to the equivalent tests in test_provider_bootstrap.py and test_provider_catalog_bootstrap.py.

Verification

• python -m pytest tests/test_discovery_bootstrap_selection.py tests/test_model_discovery.py tests/test_model_discovery_boundaries.py tests/test_provider_bootstrap.py tests/test_provider_bootstrap_boundaries.py tests/test_provider_catalog_bootstrap.py tests/test_provider_catalog_store.py -q: 182 passed.
• interrogate tests/test_discovery_bootstrap_selection.py: 100%.

Test-only change; no production code touched.

---

Generated by Claude Code

[coderabbitai]

Important

Draft PR not reviewed

Draft PRs are not automatically reviewed by default.

• Trigger a manual review

To automatically review draft PRs, update your CodeRabbit configuration:

reviews:
  auto_review:
    drafts: true

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[seonghobae]

opencode-review failure on head 57aa127 is the expected "no current-head verdict yet" gate, not a defect: the job requires an APPROVED/CHANGES_REQUESTED review from opencode-agent on this exact head SHA before it can pass, and fails closed until the async dispatch posts one. This PR was just pushed; watching for the real dispatch verdict.

---

Generated by Claude Code

[seonghobae]

Superseded by #945 (merged 0adca47), which fixes this exact same stale test with an equivalent change, opened concurrently by another session. Closing without merging to avoid duplicating already-landed work.

---

Generated by Claude Code