Trac #16474: timeouts in doctests on OS X 10.9

There are frequent timeouts when doctesting on OS X with pexpect
interfaces, in particular Singular. For example:
{{{
$ ./sage -tp src/doc/en/constructions

...
----------------------------------------------------------------------
sage -t src/doc/en/constructions/algebraic_geometry.rst  # Timed out
sage -t src/doc/en/constructions/polynomials.rst  # Timed out
----------------------------------------------------------------------
}}}
These became much more frequent by the changes in the Singular/pexpect
interface from #15631 on two buildbots:
* OSX 10.9 (Oxford)
* Ubuntu 8.04 (UW)
* This happens mostly during parallel doctesting, and disappers when the
system load is low
Other build bots do not seem to be affected, so its a likely pty kernel
bug.

URL: http://trac.sagemath.org/16474
Reported by: jhpalmieri
Ticket author(s): Volker Braun
Reviewer(s): John Palmieri

src/sage/interfaces/expect.py

@@ -437,8 +437,6 @@ def _start(self, alt_message=None, block_during_init=True):
 
         os.chdir(current_path)
         self._expect.timeout = self.__max_startup_time
-        if not self._terminal_echo:
-            self._expect.setecho(0)
 
         #self._expect.setmaxread(self.__maxread)
         self._expect.maxread = self.__maxread
@@ -451,6 +449,14 @@ def _start(self, alt_message=None, block_during_init=True):
             failed_to_start.append(self.name())
             raise RuntimeError("Unable to start %s"%self.name())
         self._expect.timeout = None
+
+        # Calling tcsetattr earlier exposes bugs in various pty
+        # implementations, see :trac:`16474`. Since we haven't
+        # **written** anything so far it is safe to wait with
+        # switching echo off until now.
+        if not self._terminal_echo:
+            self._expect.setecho(0)
+
         with gc_disabled():
             if block_during_init:
                 for X in self.__init_code: