Fix a bufferevent-related bug that killed tunneled dirserv conns

Because tunneled connections are implemented with buffervent_pair,
writing to them can cause an immediate flush.  This means that
added to them and then checking to see whether their outbuf is
empty is _not_ an adequate way to see whether you added anything.
This caused a problem in directory server connections, since they
would try spooling a little more data out, and then close the
connection if there was no queued data to send.

This fix should improve matters; it only closes the connection if
there is no more data to spool, and all of the spooling callbacks
are supposed to put the dirconn into dir_spool_none on completion.

This is bug 3814; Sebastian found it; bugfix on 0.2.3.1-alpha.

changes/bug3814

@@ -0,0 +1,4 @@
+  o Major bugfixes (bufferevents):
+    - Fix a bug where server-side tunneled bufferevent-based directory
+      streams would get closed prematurely. Fixes 3814, bugfix on
+      0.2.3.1-alpha.

src/or/directory.c

@@ -3545,8 +3545,21 @@ connection_dir_finished_flushing(dir_connection_t *conn)
       conn->_base.state = DIR_CONN_STATE_CLIENT_READING;
       return 0;
     case DIR_CONN_STATE_SERVER_WRITING:
+      if (conn->dir_spool_src != DIR_SPOOL_NONE) {
+#ifdef USE_BUFFEREVENTS
+        /* This can happen with paired bufferevents, since a paired connection
+         * can flush immediately when you write to it, making the subsequent
+         * check in connection_handle_write_cb() decide that the connection
+         * is flushed. */
+        log_debug(LD_DIRSERV, "Emptied a dirserv buffer, but still spooling.");
+#else
+        log_warn(LD_BUG, "Emptied a dirserv buffer, but it's still spooling!");
+        connection_mark_for_close(TO_CONN(conn));
+#endif
+      } else {
         log_debug(LD_DIRSERV, "Finished writing server response. Closing.");
         connection_mark_for_close(TO_CONN(conn));
+      }
       return 0;
     default:
       log_warn(LD_BUG,"called in unexpected state %d.",