fix(labs): strict numeric parsing, atomic confirm guard, test integrity

- normalize.ts num(): parseFloat truncated at the first comma, so "1,200"
  silently became 1 (1000x error) and "5,5" became 5. Now strictly matches
  either US thousands-grouping or a plain number spanning the whole string;
  anything else (incl. ambiguous "5,5") returns null instead of a wrong value.
- labs.ts confirm handler: the pending-status check ran before the request
  body was read, so two concurrent confirms could both pass it and
  double-insert. Added a guarded UPDATE ... WHERE status = 'pending' as the
  first statement inside the existing synchronous transaction; zero rows
  affected throws and the route returns 409, with the fast-path check kept
  for the common case.
- Added missing `await` on two rejects.toThrow assertions (llm.test.ts,
  extract.test.ts) that were previously resolving before the assertion
  settled.
- Bumped the 11th-failed-login rate-limit test to a 30s timeout — 10
  sequential argon2id verifies can exceed bun:test's 5s default under load.

server/test/auth.test.ts

@@ -54,7 +54,7 @@ describe("auth", () => {
     // The global window applies to everyone, including a request with the correct password.
     const blocked = await app.request("/api/login", json({ password: "hunter2hunter2" }));
     expect(blocked.status).toBe(429);
-  });
+  }, 30000); // 10 sequential argon2id verifies can exceed the 5s default timeout
 
   test("unknown /api/* returns 404 when authed, 401 when unauthed", async () => {
     const app = makeApp();

server/test/extract.test.ts

@@ -41,6 +41,6 @@ describe("extractFromText", () => {
   test("malformed LLM output → llm_error, not a crash", async () => {
     const mock = (async () => new Response(JSON.stringify({ choices: [{ message: { content: '{"markers": "not an array"}' } }] }), { status: 200 })) as unknown as typeof fetch;
     const dir = mkdtempSync(join(tmpdir(), "helios-"));
-    expect(extractFromText({ db: openDb(dir), key: loadOrCreateKey(dir), fetchImpl: mock }, "x")).rejects.toThrow(/llm_error/);
+    await expect(extractFromText({ db: openDb(dir), key: loadOrCreateKey(dir), fetchImpl: mock }, "x")).rejects.toThrow(/llm_error/);
   });
 });

server/test/labs-confirm.test.ts

@@ -64,6 +64,13 @@ describe("draft review", () => {
       body: JSON.stringify({ collectedDate: "2026-01-15", labName: null, markers: [{ panel: null, name: "X", value: "1", unit: null, referenceRange: null, flagged: false }] }),
     });
     expect(again.status).toBe(409);
+
+    // Second (rejected) confirm must not have inserted a second draw or any
+    // extra biomarker rows — exactly one draw, one set of biomarkers.
+    const drawsAfter = await db.select().from(labDraws);
+    expect(drawsAfter).toHaveLength(1);
+    const rowsAfter = await db.select().from(biomarkers);
+    expect(rowsAfter).toHaveLength(2);
   });
 
   test("discard marks draft discarded", async () => {

server/test/llm.test.ts

@@ -46,6 +46,6 @@ describe("chatJSON", () => {
 
   test("non-2xx throws llm_error", async () => {
     const mock = (async () => new Response("nope", { status: 401 })) as unknown as typeof fetch;
-    expect(chatJSON(deps(mock), { system: "s", user: "u" })).rejects.toThrow(/llm_error/);
+    await expect(chatJSON(deps(mock), { system: "s", user: "u" })).rejects.toThrow(/llm_error/);
   });
 });

server/test/normalize.test.ts

@@ -61,4 +61,21 @@ describe("normalizeMarker", () => {
     const n = normalizeMarker({ name: "DHEA-S", value: "250", unit: "ug/dL" });
     expect(n.valueCanonical).toBeCloseTo(6.78, 1); // 250 * 0.02713
   });
+  test("thousands-grouped value parses as 1200, not 1 (parseFloat truncation bug)", () => {
+    const n = normalizeMarker({ name: "Vitamin B12", value: "1,200", unit: "pg/mL" });
+    expect(n.valueNum).toBe(1200);
+    expect(n.analyteKey).toBe("vitamin_b12");
+    expect(n.valueCanonical).toBeCloseTo(885.35, 1);
+    expect(n.canonicalUnit).toBe("pmol/L");
+  });
+  test("ambiguous European-decimal-looking value is left unmapped, not silently wrong", () => {
+    const n = normalizeMarker({ name: "Glucose", value: "5,5", unit: "mg/dL" });
+    expect(n.valueNum).toBeNull();
+    expect(n.valueCanonical).toBeNull();
+    expect(n.value).toBe("5,5"); // raw value preserved
+  });
+  test("comparator value still parses after strict-parsing rewrite", () => {
+    const n = normalizeMarker({ name: "hs-CRP", value: "<0.3", unit: "mg/L" });
+    expect(n.valueNum).toBeCloseTo(0.3);
+  });
 });