fix(test): accept endpoint kwarg in local-appdata trust test stub (#5879)

[CrysisDeu]

Summary

Main unblocker: one-line test fix. TestWindowsCliStore::test_local_appdata_store_token_is_trusted_without_arn fails on all backend shards of main's own CI (and on every PR merge ref) with assert None is not None.

Closes #5879

Root cause

PR #4545 changed _post in src/kiro_crew/dashboard/handlers/kiro_usage_api.py to take a keyword-only parameter:

def _post(token: str, target: str, payload: dict, *, endpoint: str | None = None) -> _Resp:

Both production call sites (:598, :901) now pass endpoint=. The failing test's stub at test/test_kiro_usage_api.py:505 was declared def fake_post(token, target, payload) without **_kwargs, so the side_effect raises TypeError inside fetch_usage_limits, which degrades to returning None — hence the assertion failure.

Fix

Add **_kwargs to the stub's signature, byte-matching the passing sibling stub at :466. Grep confirms every other _post stub in the file (fake_post × 7, recording_post × 3) already accepts **_kwargs; line 505 was the only miss.

Deliberately minimal diff (1 line): this unblocks PR Readiness repo-wide. The Frontend Tests shard 3 red on main is a separate issue (#5529) and is not touched here.

Verification

• Red→green proven locally on main head 6ed29f5cb: the test fails before the change (assert None is not None) and passes after; full test/test_kiro_usage_api.py = 88 passed.
• Local gates: isort / flake8 / mypy clean; black diff-gate passed; full pytest = 68281 passed, 15 failures all in files untouched by this diff (host-env classes: test_xdist_host_budget, test_artifact_source, test_host_isolation_floor, test_trust_reads — byte-identical to main).
• Pre-push model-pinned reviews: GPT lane (gpt-5.6-sol) PASS, zero findings. Opus lane (claude-opus-5) PASS, zero blocking; one advisory noted below.

Advisory disposition

Opus advisory: the stub swallows endpoint into **_kwargs, so the test cannot observe whether the regional endpoint is threaded to _TARGET_GET_USAGE. Capturing and asserting it is real coverage of #4545's behavior but widens the diff beyond a one-line main unblocker — deferred, not adopted here.

[github-actions]

Design Review (Fable 5) — ✅ PASS

Design-level review of 169347f5d3501802e94ecea23167a1545392defc — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

The diff is a one-line test stub fix, exactly as described: line 505's fake_post was the sole stub missing **_kwargs, and the change aligns it with every sibling in the file. The description matches the diff bidirectionally, the fix addresses the actual root cause (stub signature drift after #4545's keyword-only parameter), and the deferred coverage advisory is a reasonable scoping call for a CI unblocker.

Design-Verdict: PASS

One-line stub-signature fix that unblocks CI at its root cause; minimal, correct scope, no design surface touched.

[DESIGN-REVIEWED] 169347f

[github-actions]

GPT 5.6 Review — ✅ no blocking findings

GPT 5.6 completed its review of 169347f5d3501802e94ecea23167a1545392defc and found no blocking issues.

This comment is updated in place on each push.

Review details

No findings.
[GPT-REVIEWED] 169347f

False positive or not applicable? A repository writer can comment:
/ai-review override gpt 169347f5d3501802e94ecea23167a1545392defc: <one-sentence reason>

[github-actions]

Opus 4.8 Review — ✅ no blocking findings

Reviewed 169347f5d3501802e94ecea23167a1545392defc — this comment is updated in place on each push.

Review details

No findings.

[OPUS-REVIEWED] 169347f

Verdict parsed from the review's SHA-scoped output markers for commit 169347f5d3501802e94ecea23167a1545392defc.

False positive or not applicable? A repository writer can comment:
/ai-review override fable 169347f5d3501802e94ecea23167a1545392defc: <one-sentence reason>

[CrysisDeu]

Status: review-ready — the 5 red checks are inherited from main, not this PR's.

Evidence for the maintainer:

• The 5 reds are Backend Tests (Windows) (4), (3.10, 4), (3.12, 4) → Coverage Gate → PR Readiness. Each shard-4 job fails on exactly one test: test/test_trust_reads.py::TestIsReadOnlyBash::test_a_word_bash_deletes_cannot_forge_a_read_mode (AssertionError: wc -l < f) — issue main red: #5521 raw-scan refuses all input redirects, contradicting #5541 pinned redirect-read allowance #5846, red on main itself, zero overlap with this PR's one-line diff in test/test_kiro_usage_api.py.
• The shard-2 jobs (where the test this PR fixes lives) are all green on this PR's merge ref, proving the fix works.
• Everything this PR owns is green (58/63), MERGEABLE, and all AI review lanes settled PASS on head 169347f5d (Design PASS, GPT no blocking findings, Opus no blocking findings).

Cross-deadlock note: #5880 fixes #5846 (trust_reads) and inherits this PR's usage-api red on its shard-2 jobs; this PR inherits #5880's trust_reads red on shard-4. Neither can reach a fully green board until one of them merges — merging either one first (with its known inherited red), then rebasing the other, greens both and un-blocks PR Readiness repo-wide.

[bolichen97]

Full-diff overlap audit (current head 169347f5d3501802e94ecea23167a1545392defc): this PR's entire diff is the one-line test/test_kiro_usage_api.py change from fake_post(token, target, payload) to fake_post(token, target, payload, **_kwargs). Merged #3410 (ff15bbe14c6e45f1e1c9b218106cf77a49b5e6ad) carries the byte-identical hunk. The path-filtered stable patch-id is identical (b53e2b53b67cdb0cb3c45c87a5953a507351fb41), and the resulting file blob already matches main.

Closing as exactly covered by merged #3410.