EngineClient: skip closing eof on non-existing streams

We close stdin when clush's stdin closes as well as when remote workers
exit.

Fixes this kind of trace:
$ echo | clush -w foobar[0-11] -d echo
...
WARNING:ClusterShell.Worker.EngineClient:<ClusterShell.Worker.Ssh.SshClient instance at 0x28eaab8>: [Errno 32] Broken pipe
...
foobar0: ssh: Could not resolve hostname foobar0: Name or service not known
...
clush: foobar0: exited with exit code 255
Error in sys.excepthook:
Traceback (most recent call last):
  File "/usr/lib/python2.7/site-packages/ClusterShell/CLI/Clush.py", line 751, in clush_excepthook
    raise exp
KeyError: 'stdin'

Original exception was:
Traceback (most recent call last):
  File "/usr/bin/clush", line 10, in <module>
    main()
  File "/usr/lib/python2.7/site-packages/ClusterShell/CLI/Clush.py", line 1069, in main
    options.remote != 'no')
  File "/usr/lib/python2.7/site-packages/ClusterShell/CLI/Clush.py", line 667, in run_command
    task.resume()
  File "/usr/lib/python2.7/site-packages/ClusterShell/Task.py", line 799, in resume
    self._resume()
  File "/usr/lib/python2.7/site-packages/ClusterShell/Task.py", line 762, in _resume
    self._run(self.timeout)
  File "/usr/lib/python2.7/site-packages/ClusterShell/Task.py", line 407, in _run
    self._engine.run(timeout)
  File "/usr/lib/python2.7/site-packages/ClusterShell/Engine/Engine.py", line 700, in run
    self.runloop(timeout)
  File "/usr/lib/python2.7/site-packages/ClusterShell/Engine/EPoll.py", line 165, in runloop
    client._handle_read(sname)
  File "/usr/lib/python2.7/site-packages/ClusterShell/Worker/EngineClient.py", line 520, in _handle_read
    self.eh.ev_msg(self, pmsg.get())
  File "/usr/lib/python2.7/site-packages/ClusterShell/CLI/Clush.py", line 85, in ev_msg
    self.master_worker.set_write_eof()
  File "/usr/lib/python2.7/site-packages/ClusterShell/Worker/Exec.py", line 365, in set_write_eof
    client._set_write_eof(sname or self.SNAME_STDIN)
  File "/usr/lib/python2.7/site-packages/ClusterShell/Worker/EngineClient.py", line 415, in _set_write_eof
    wfile = self.streams[sname]
KeyError: 'stdin'

Closes #320.

Change-Id: I877e036548dd95abda4494e420ab6c9bc5138bff

lib/ClusterShell/Worker/EngineClient.py

@@ -418,6 +418,11 @@ def _write(self, sname, buf):
 
     def _set_write_eof(self, sname):
         """Set EOF on specific writable stream."""
+        if sname not in self.streams:
+            LOGGER.debug("stream %s was already closed on client %s, skipping",
+                         sname, self.key)
+            return
+
         wfile = self.streams[sname]
         wfile.eof = True
         if self._engine and wfile.fd and not wfile.wbuf: