[gm.py] Fix gm.py to work on Windows

gm.py has been careful to only use the "pty" module's functionality
on Linux, but as it turns out, the module is *so* strongly specific
to Linux that even importing it fails on Windows. Making the import
of "pty" conditional makes gm.py work on Windows.

Change-Id: I0c1fb8a9a0299fde50e252337551d9395039f14d
Reviewed-on: https://chromium-review.googlesource.com/c/1382738
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#58335}

tools/dev/gm.py

@@ -22,11 +22,14 @@ All arguments are optional. Most combinations should work, e.g.:
 from __future__ import print_function
 import errno
 import os
-import pty
 import re
 import subprocess
 import sys
 
+USE_PTY = "linux" in sys.platform
+if USE_PTY:
+  import pty
+
 BUILD_TARGETS_TEST = ["d8", "cctest", "unittests"]
 BUILD_TARGETS_ALL = ["all"]
 
@@ -258,7 +261,7 @@ class Config(object):
     targets = " ".join(self.targets)
     # The implementation of mksnapshot failure detection relies on
     # the "pty" module and GDB presence, so skip it on non-Linux.
-    if "linux" not in sys.platform:
+    if not USE_PTY:
       return _Call("autoninja -C %s %s" % (path, targets))
 
     return_code, output = _CallWithOutput("autoninja -C %s %s" %