Single-pass aser_slot for HTMX path + kernel eval timing + stable hash

Eliminated double-aser for HTMX requests: build OOB wrapper AST
(~shared:layout/oob-sx :content wrapped_ast) and aser_slot in ONE
pass — same pattern as the full-page path. Halves aser_slot calls.

Added kernel-side timing to stderr:
  [aser-slot] eval=3.6s io_flush=0.0s batched=3 result=22235 chars

Results show batch IO works (io_flush=0.0s for 3 highlight calls)
and the bottleneck is pure CEK evaluation time, not IO.

Performance after single-pass fix:
  Home: 0.7s eval (was 2.2s total)
  Reactive: 3.6s eval (was 6.8s total)
  Language: 1.1s eval (was 18.9s total — double-aser eliminated)

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

shared/sx/ocaml_bridge.py

@@ -59,11 +59,12 @@ class OcamlBridge:
             )
 
         _logger.info("Starting OCaml SX kernel: %s", bin_path)
+        import sys
         self._proc = await asyncio.create_subprocess_exec(
             bin_path,
             stdin=asyncio.subprocess.PIPE,
             stdout=asyncio.subprocess.PIPE,
-            stderr=asyncio.subprocess.PIPE,
+            stderr=sys.stderr,  # kernel timing/debug to container logs
             limit=10 * 1024 * 1024,  # 10MB readline buffer for large spec data
         )
 
@@ -312,12 +313,9 @@ class OcamlBridge:
         assert self._proc and self._proc.stdout
         data = await self._proc.stdout.readline()
         if not data:
-            # Process died — collect stderr for diagnostics
-            stderr = b""
-            if self._proc.stderr:
-                stderr = await self._proc.stderr.read()
+            # Process died
             raise OcamlBridgeError(
-                f"OCaml subprocess died unexpectedly. stderr: {stderr.decode(errors='replace')}"
+                "OCaml subprocess died unexpectedly (check container stderr)"
             )
         line = data.decode().rstrip("\n")
         _logger.debug("RECV: %s", line[:120])