Enable pre-warm cache + faster Playwright config

Restore page pre-warming at HTTP server startup (was skipped) and
increase render workers from 2→4. Tighten Playwright timeouts and
run 3 workers in parallel for faster test runs.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

hosts/ocaml/bin/sx_server.ml

@@ -2044,8 +2044,11 @@ let http_mode port =
                       "/sx/(language)"; "/sx/(applications)";
                       "/sx/(geography.(reactive))"; "/sx/(geography.(hypermedia))";
                       ] in
-  ignore cache_response;
-  Printf.eprintf "[sx-http] Skipping pre-warm (render on demand)\n%!";
+  let t_warm = Unix.gettimeofday () in
+  List.iter cache_response _warmup_paths;
+  let n_cached = Hashtbl.length response_cache in
+  Printf.eprintf "[sx-http] Pre-warmed %d pages in %.3fs\n%!"
+    n_cached (Unix.gettimeofday () -. t_warm);
 
   (* Write full response to a socket *)
   let write_response client response =
@@ -2080,7 +2083,7 @@ let http_mode port =
      - Cache misses: queued to render workers (domain pool)
      - Never blocks on rendering — accept loop always responsive *)
 
-  let n_workers = max 2 (Domain.recommended_domain_count ()) in
+  let n_workers = max 4 (Domain.recommended_domain_count ()) in
 
   (* Render queue: for cache misses that need full page render *)
   let render_queue : (Unix.file_descr * string) list ref = ref [] in