Add Mirror List Command Tests

Done. Both review comments on PR #1490 are addressed.

Bottom line

Addressed the requested PR review comment (and the adjacent one) on cmd/entire/cli/repo_mirror.go.

What changed:

• repo_mirror_test.go — Added TestRepoMirrorList_ShowAvailableRouting, a command-level test that executes repo mirror list end-to-end against an httptest fake control-plane and locks in:
  • --show-available → routes to GET /api/v1/mirrors/available, renders the ACCESS/STATUS columns (not CLONE URL), and prints the "Listing repos you could mirror" banner.
  • default → routes to GET /api/v1/mirrors with the clone-URL view + "Listing mirrors on" banner.
  • --owner flows into the query on both paths.
  • --cluster/--provider apply to /mirrors but are correctly ignored by --show-available.
• corecmd.go — Added a small activeCoreClient package seam (production = coreapi.New) so runCore can be pointed at a test server without the auth/context/TLS stack. (The original review comment, r3449752471.)
• repo_mirror.go — Fixed the availableMirrorColumns doc comment to use the real entire repo mirror create/list command names instead of mirror create/mirror list. (The adjacent comment, r3449752460.)

Heads-up — repo footgun I hit: mise run lint runs golangci-lint --fix, and a stale lint cache made nolintlint think the //nolint:ireturn directives in checkpoint/open.go and strategy/manual_commit.go were unused, auto-stripping them from those two unrelated files. I restored the directives (they now match HEAD) and confirmed it was a cache artifact — golangci-lint cache clean then a full run reports 0 issues. If you mise run lint with a stale cache it may strip them again; clear the cache first.

Tests pass; lint is clean. Not committed — yours to push.

Pushed.

• Rebased onto remote (it had a "Comment nitpicks" commit that already made the same doc fix — I dropped my redundant wording and kept the test + seam).
• f388237b1 now on origin/mirror-list-show-available. Build + test green.

Fixed and pushed. I did not mark the finding resolved on the trail — you only asked me to look, so I'll leave that to you.

Bottom line

One open high-severity finding on trail #631 — a real data race in my test helper. Fixed and pushed.

The finding: serveMirrorList wrote recordedRequest from the HTTP handler goroutine and read it from the test goroutine with no happens-before edge — -race would fail it. (Same gotcha the existing TestBearerOnlySource_NoCookieOnTheWire documents.)

The fix (8292221f9): Deliver the recorded request over a buffered channel; the test receives from it after the command runs, which is the synchronization edge. Verified with go test -race — all 5 subtests pass clean. Lint/gofmt clean.

To close it out: the finding (019eedf8-…) is still open — run when ready: