perf: dedicated 500-worker thread pool for ubridge batch I/O

Replace the default asyncio executor (capped at ~32 threads) with a
dedicated ThreadPoolExecutor sized for large-topology parallelism.
When 500 nodes each call _connect_nio, up to 500 OS threads can now
send blocking ubridge commands in parallel — no longer serialised by
either the event loop or a small thread pool.

- ubridge_hypervisor: module-level _ubridge_sync_pool (max_workers=500)
- docker_vm._connect_nio: dispatches to the dedicated pool instead of
  the default executor

gns3server/compute/docker/docker_vm.py

@@ -38,6 +38,7 @@ from gns3server.utils.hostname import is_rfc1123_hostname_valid
 from gns3server.utils import macaddress_to_int, int_to_macaddress
 
 from gns3server.compute.ubridge.ubridge_error import UbridgeError, UbridgeNamespaceError
+from gns3server.compute.ubridge.ubridge_hypervisor import _ubridge_sync_pool
 from ..base_node import BaseNode
 
 from ..adapters.ethernet_adapter import EthernetAdapter
@@ -1242,7 +1243,7 @@ class DockerVM(BaseNode):
         async with getattr(self, lock_name):
             loop = asyncio.get_running_loop()
             await loop.run_in_executor(
-                None,  # default thread-pool executor
+                _ubridge_sync_pool,  # dedicated 500-worker thread pool
                 self._ubridge_hypervisor.send_batch_sync,
                 commands,
             )

gns3server/compute/ubridge/ubridge_hypervisor.py

@@ -19,12 +19,23 @@ import time
 import logging
 import asyncio
 import threading
+import concurrent.futures
 
 from gns3server.utils.asyncio import locking
 from .ubridge_error import UbridgeError
 
 log = logging.getLogger(__name__)
 
+# Dedicated thread pool for blocking ubridge socket I/O.  Every node gets
+# its own ubridge process + socket, so N nodes can send commands in true
+# OS-thread parallelism.  The default asyncio executor caps at ~32 threads;
+# sizing for the large-topology case (2500+ links → thousands of NIO add
+# calls across hundreds of nodes).
+_ubridge_sync_pool = concurrent.futures.ThreadPoolExecutor(
+    max_workers=500,
+    thread_name_prefix="ubridge-sync",
+)
+
 
 class UBridgeHypervisor: