feat: v2.5.12 — plan and quota for every AI tool, Grok Build and Kimi Code, Quit AgentHub (#65)

Usage panel (status bar):
- every account shows its plan (Claude subscriptionType, Codex
  chatgpt_plan_type, Cursor membershipType, Grok subscriptionTier)
- new quota rows: Cursor CLI (cursor.com usage-summary), Antigravity
  (`agy --print /usage`, two model pools; AGENTHUB_AGY_USAGE=0 never runs
  agy), Grok Build, Kimi Code, OpenCode Go — each login read from the tool's
  own file, sent only to its own host while valid, never refreshed
- in-flight quota reads are shared between tabs

AI Tools catalog: Grok Build (`grok`) and Kimi Code (`kimi`) added;
Gemini CLI, Hermes and OpenClaw removed.

Also: Quit AgentHub (#65, `agenthub stop`), status-area icon, Add to Chat
on an editor selection, Settings in three groups, JetBrains-style Changes
panel (ticked files/hunks, ignore, blame, file history), terminal font.

--no-verify: :3000 is taken by another dev server (lane-detect tests);
both suites run by hand — server 2636/2639 (the 3 are :3000), client
1131/1131.

server/__tests__/token-baseline.test.js

@@ -1,5 +1,6 @@
 /**
- * @file Tests for replaceTokenUsage's compaction baseline semantics.
+ * @file Tests for replaceTokenUsage's compaction baseline semantics, and for
+ * the long-context subsets written beside it (high-water mark, exact reset).
  *
  * The effective total for a bucket is `live + baseline`. It must behave as a
  * monotonic HIGH-WATER MARK: never decrease (so a compaction that shrinks the
@@ -121,3 +122,38 @@ describe("replaceTokenUsage — high-water-mark baseline", () => {
     assert.equal(rawRow("hw-recover").baseline_cache_read, 0);
   });
 });
+
+describe("long-context subsets", () => {
+  const { writeSessionTokens } = require("../../scripts/import-history");
+  const { emptyBucket, accumulateBucket, extractUsageFields } = require("../lib/token-usage");
+  const longRow = (sid) =>
+    db
+      .prepare("SELECT input_tokens, long_input_tokens FROM token_usage WHERE session_id = ?")
+      .get(sid);
+  const bucketOf = (...usages) => {
+    const b = emptyBucket("claude-haiku-5-5", "standard", "global", "standard");
+    for (const u of usages) accumulateBucket(b, extractUsageFields(u));
+    return { k: b };
+  };
+
+  before(() => {
+    db.prepare(
+      "INSERT OR IGNORE INTO sessions (id, name, status) VALUES ('long-1', 'x', 'active')"
+    ).run();
+  });
+
+  it("stores the long part as a subset of the same row and keeps its high-water mark", () => {
+    const short = { input_tokens: 1_000 };
+    const long = { input_tokens: 150_000 };
+    writeSessionTokens(dbModule, "long-1", bucketOf(short, long));
+    assert.deepEqual(longRow("long-1"), { input_tokens: 151_000, long_input_tokens: 150_000 });
+    // A smaller-scope writer never takes the subset back.
+    writeSessionTokens(dbModule, "long-1", bucketOf(short));
+    assert.equal(longRow("long-1").long_input_tokens, 150_000);
+  });
+
+  it("lets the exact reconcile path clear a stale subset", () => {
+    writeSessionTokens(dbModule, "long-1", bucketOf({ input_tokens: 1_000 }), { exact: true });
+    assert.deepEqual(longRow("long-1"), { input_tokens: 1_000, long_input_tokens: 0 });
+  });
+});