Stabilize launch-test timeouts to prevent CI false negatives

[Copilot]

CI / ROS iron (main) (pull_request) failed due to a launch test timeout (test_test_zmq.py) that did not emit a result file before CTest’s timeout window. The failure mode indicates timing sensitivity in CI rather than a deterministic logic regression.

• Root cause

  • Launch-test timeout configuration in CMakeLists.txt was too aggressive for slower CI runners.
  • Timeout-triggered termination caused missing xUnit output (test_test_zmq.py.xunit.missing_result).

• Change made

  • Introduced a shared, configurable launch-test timeout in BUILD_TESTING scope:
    • NETWORK_BRIDGE_LAUNCH_TEST_TIMEOUT (default: 30)
  • Replaced hardcoded launch test timeouts for existing launch tests to use the shared value.

• Why this shape

  • Keeps behavior centralized and tunable.
  • Applies consistently across launch tests without touching test logic.

set(NETWORK_BRIDGE_LAUNCH_TEST_TIMEOUT 30 CACHE STRING "Timeout in seconds for launch tests")

add_launch_test(
  test/test_udp.py
  TIMEOUT ${NETWORK_BRIDGE_LAUNCH_TEST_TIMEOUT}
)
add_launch_test(
  test/test_tcp.py
  TIMEOUT ${NETWORK_BRIDGE_LAUNCH_TEST_TIMEOUT}
)