fix(kernel-node-runtime): increase ACK timeout for multi-peer reconnection test

The "handles multiple simultaneous reconnections to different peers"
test has kernel1 redeem URLs on two peers that restart sequentially
through a relay. The default ackTimeoutMs of 2s gives only 8s for
URL redemption, which is too tight when two kernels are reconnecting
in CI. Use 5s (→ 20s redemption window) for the initiating kernel.

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

Author: sirtimid

packages/kernel-node-runtime/test/e2e/remote-comms.test.ts

@@ -651,10 +651,15 @@ describe.sequential('Remote Communications E2E', () => {
         });
         let kernel3: Kernel | undefined;
 
+        // Use a longer ACK timeout for the initiating kernel — it must redeem
+        // URLs on two peers that restart sequentially through the relay, so the
+        // default 2s × (3+1) = 8s redemption window is too tight for CI.
+        const multiPeerBackoff = { ...testBackoffOptions, ackTimeoutMs: 5_000 };
+
         try {
           await kernel1.initRemoteComms({
             relays: testRelays,
-            ...testBackoffOptions,
+            ...multiPeerBackoff,
           });
           await kernel2.initRemoteComms({
             relays: testRelays,
@@ -757,7 +762,7 @@ describe.sequential('Remote Communications E2E', () => {
           }
         }
       },
-      NETWORK_TIMEOUT * 3,
+      NETWORK_TIMEOUT * 4,
     );
   });