yakut call: Spurious redundant transport errors on remote end

yakut call will shut down the local node immediately after the response is received. If the transport is redundant and one of the inferiors delivered the response faster than the other, the transport connections may be closed by the client while the remote end (the server) is still trying to send the response, causing the slower inferiors to fail. This obviously does not affect the connectivity since the primary purpose of the redundant transport is to be resilient against these types of errors, but it does result in rather useless error reports on the server side like this:

ERROR
  pycyphal.transport.redundant._session._output:_output.py:335
    RedundantOutputSession(OutputSessionSpecifier(data_specifier=ServiceDataSpecifier(service_id=222, role=<Role.RESPONSE: 2>), remote_node_id=88), PayloadMetadata(extent_bytes=64)):
      Inferior UDPOutputSession(OutputSessionSpecifier(data_specifier=ServiceDataSpecifier(service_id=222, role=<Role.RESPONSE: 2>), remote_node_id=88), PayloadMetadata(extent_bytes=64)) failed:
        ConnectionRefusedError: [Errno 111] Connection refused

This particular error is occasionally generated by the test suite when testing the heterogeneous UDP+serial configuration (slightly related https://github.com/OpenCyphal/pycyphal/issues/222).

Perhaps the "call" command should wait briefly before shutting down the connections. This is a low-priority problem though.

Start with the yakut call command and the heterogeneous UDP+serial transport test configuration mentioned in the issue. Reproduce the redundant-response shutdown behavior and inspect the reported pycyphal.transport.redundant._session._output failure; done means the remote end no longer emits spurious redundant transport errors after the call response is received.