feat: add the two-tier dashboard (task 0017)

Wire the bare `gitea-axi` home command to a two-tier repository
dashboard (ADR 0012): the short tier shows up to 3 open issues and 3
open PRs with the client-side `review` decision, and `--full` shows the
20-row open-PR table plus open issue counts grouped by label, aggregated
across every page of open issues up to the 1000-issue cap with a `+`
suffix when capped. Empty states render the raw `prs: 0 open` /
`issues: 0 open` strings; issue fetches pass `type=issues`; outside a
Gitea repo the dashboard errors with REPO_NOT_FOUND.

`paginate.ts` now reports whether pagination stopped at the cap, which
drives the label-count `+` suffix.

test/e2e/dashboard.test.ts

@@ -0,0 +1,79 @@
+import { beforeAll, describe, expect, it } from "vitest";
+import { runCliTest } from "../harness.js";
+import { provisionInstance, seedBranch, type E2EInstance } from "./provision.js";
+
+/**
+ * The end-to-end tier for the dashboard. The short and full tiers read the live
+ * issue-list and PR-list responses, then compute two fields the fixture server
+ * can only stub: the per-PR `review` (folded from a separate reviews fetch) and
+ * the full-tier label aggregation (issue counts grouped by label). Those live
+ * response shapes and derived fields are exactly what fixtures cannot attest to,
+ * so here both tiers run against a live, disposable Gitea instance. Unlike the
+ * search tier, the dashboard hits immediately-consistent list endpoints, so no
+ * indexer polling is needed — assertions run directly after provisioning.
+ */
+const E2E_URL = process.env.GITEA_AXI_E2E_URL;
+
+describe.skipIf(!E2E_URL)("end-to-end: dashboard", () => {
+  let instance: E2EInstance;
+  const branch = "e2e-dashboard-branch";
+
+  function env(overrides: Record<string, string> = {}): Record<string, string> {
+    return {
+      GITEA_AXI_API_URL: instance.baseUrl,
+      GITEA_AXI_TOKEN: instance.token,
+      GITEA_AXI_REPO: `${instance.owner}/${instance.repo}`,
+      ...overrides,
+    };
+  }
+
+  beforeAll(async () => {
+    instance = await provisionInstance(E2E_URL!);
+    // A fresh repo has no pull requests; seed a branch with a diff and open one
+    // through the CLI dogfood path so the dashboard's PR block has a real row.
+    await seedBranch(instance, branch);
+    const created = await runCliTest(
+      [
+        "pr", "create",
+        "--title", "E2E dashboard pull request",
+        "--head", branch,
+        "--base", "main",
+      ],
+      { env: env() },
+    );
+    expect(created.exitCode).toBe(0);
+  }, 150_000);
+
+  it("renders the bare dashboard with live issue and PR list shapes and a computed review", async () => {
+    const { stdout, exitCode } = await runCliTest([], { env: env() });
+
+    expect(exitCode).toBe(0);
+    expect(stdout).toContain(`repo: ${instance.owner}/${instance.repo}`);
+
+    // The issue block is the short-tier list shape, populated from the live
+    // issue-list response — one of the seeded open titles must appear.
+    expect(stdout).toMatch(/^issues\[\d+\]\{number,title,state,author\}:$/m);
+    expect(stdout).toContain(instance.openTitles[0]!);
+
+    // The PR block carries the client-side `review` field. The seeded PR has no
+    // reviews yet, so it computes `required`; the union guards against the value
+    // legitimately being an approval/change-request in some run.
+    expect(stdout).toMatch(/^prs\[\d+\]\{number,title,author,review\}:$/m);
+    expect(stdout).toMatch(/,(approved|changes_requested|required)$/m);
+  });
+
+  it("renders the --full PR table and label-aggregation against live responses", async () => {
+    const { stdout, exitCode } = await runCliTest(["--full"], { env: env() });
+
+    expect(exitCode).toBe(0);
+
+    // The full tier renders the open-PR table with a labels column and a count
+    // line, both derived from the live list response.
+    expect(stdout).toMatch(/^prs\[\d+\]\{number,title,author,labels,review\}:$/m);
+    expect(stdout).toMatch(/^count: \d+ of \d+ total$/m);
+
+    // The full-tier issues block is a label->count record. The seeded open
+    // issues are unlabeled, so an `unlabeled` bucket must be present.
+    expect(stdout).toMatch(/^ {2}unlabeled: \d+$/m);
+  });
+});