doctest/control.py: add default for die_timeout

This was missing in the previous commit.

Also, tune up the timeout in the corresponding test in doctest/test.py.
This prevents a doctest failure when pytest is installed (since then
running `sage -t` will take ~ 3s after the doctest was actually run.

If #33531 is merged, the way we run `sage -t` here (using
`--optional=sage`) avoids the pytest overhead and the timeout can be
tuned down again.

src/sage/doctest/control.py

@@ -100,6 +100,7 @@ def __init__(self, **kwds):
         self.nthreads = 1
         self.serial = False
         self.timeout = -1
+        self.die_timeout = -1
         self.all = False
         self.logfile = None
         self.long = False

src/sage/doctest/test.py

@@ -237,7 +237,7 @@
     sage: kwds2 = deepcopy(kwds)
     sage: kwds2['env']['DOCTEST_TEST_PID_FILE'] = F  # Doctester will write its PID in this file
     sage: subprocess.call(["sage", "-tp", "1000000", "--timeout=120",  # long time
-    ....:      "--die_timeout=4", "--optional=sage",
+    ....:      "--die_timeout=10", "--optional=sage",
     ....:      "--warn-long", "0", "99seconds.rst", "interrupt_diehard.rst"], **kwds2)
     Running doctests...
     Doctesting 2 files using 1000000 threads...
@@ -251,13 +251,13 @@
 
 Even though the doctester master process has exited, the child process
 is still alive, but it should be killed automatically
-after the `die_timeout` given above (4 seconds)::
+after the `die_timeout` given above (10 seconds)::
 
     sage: pid = int(open(F).read())    # long time
     sage: time.sleep(2)                # long time
     sage: os.kill(pid, signal.SIGQUIT) # long time; 2 seconds passed => still alive
-    sage: time.sleep(2)                # long time
-    sage: os.kill(pid, signal.SIGQUIT) # long time; 4 seconds passed => dead
+    sage: time.sleep(8)                # long time
+    sage: os.kill(pid, signal.SIGQUIT) # long time; 10 seconds passed => dead
     Traceback (most recent call last):
     ...
     ProcessLookupError: ...