test(waterdata): close server-leak window in pool-timeout regression test

The ThreadingHTTPServer thread was started before the try block, so a
setup failure between thread.start() and the try (monkeypatch, decorator
construction) would leak the serve_forever thread and its bound socket —
the finally never ran. Move the try up to immediately after the thread
starts so teardown always fires. (server.shutdown() stays safe: it's only
reachable once serve_forever is running.)

Surfaced by the code-review sweep. No production change.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Author: thodson-usgs

tests/waterdata_chunking_test.py

@@ -1583,23 +1583,25 @@ def log_message(self, *args):  # keep pytest output clean
     server = http.server.ThreadingHTTPServer(("127.0.0.1", 0), _SlowHandler)
     thread = threading.Thread(target=server.serve_forever, daemon=True)
     thread.start()
-    url = f"http://127.0.0.1:{server.server_address[1]}/"
+    # Everything past the thread start is in the try, so any setup failure
+    # (monkeypatch, decorator construction) still tears the server down.
+    try:
+        url = f"http://127.0.0.1:{server.server_address[1]}/"
 
-    # Scale the production pool timeout (see ``HTTPX_DEFAULTS``) down to
-    # 0.2 s so the pre-semaphore failure mode reproduces in test time.
-    monkeypatch.setitem(
-        HTTPX_DEFAULTS, "timeout", httpx.Timeout(5.0, connect=1.0, pool=0.2)
-    )
+        # Scale the production pool timeout (see ``HTTPX_DEFAULTS``) down to
+        # 0.2 s so the pre-semaphore failure mode reproduces in test time.
+        monkeypatch.setitem(
+            HTTPX_DEFAULTS, "timeout", httpx.Timeout(5.0, connect=1.0, pool=0.2)
+        )
 
-    async def fetch_async(args):
-        client = get_active_client()
-        assert client is not None, "sub-request must use the shared client"
-        resp = await client.get(url)
-        assert resp.status_code == 200
-        return pd.DataFrame({"id": [_atom_id(args)]}), resp
+        async def fetch_async(args):
+            client = get_active_client()
+            assert client is not None, "sub-request must use the shared client"
+            resp = await client.get(url)
+            assert resp.status_code == 200
+            return pd.DataFrame({"id": [_atom_id(args)]}), resp
 
-    sites = _EIGHT_SINGLETON_SITES[:4]  # 2 in flight + 2 queued, 2 waves
-    try:
+        sites = _EIGHT_SINGLETON_SITES[:4]  # 2 in flight + 2 queued, 2 waves
         fetch = _async_chunked_fetch(monkeypatch, fetch_async, max_concurrent=2)
         df, _ = fetch({"sites": sites})
         assert len(df) == len(sites)