Fix TestMiniDumpNew.py test for Python 2/3 issue

[amccarth]

Summary

On Windows, where we use Python 3 for testing, we have to be more explicit about converting between binary and string representations. I believe this should still work for Python 2, but I don't have a convenient way to try it out.

lldb/trunk/packages/Python/lldbsuite/test/functionalities/postmortem/minidump-new/TestMiniDumpNew.py

def test_deeper_stack_in_minidump(self):
                                   self._linux_x86_64_not_crashed_pid)
 
     def do_change_pid_in_minidump(self, core, newcore, offset, oldpid, newpid):
         """ This assumes that the minidump is breakpad generated on Linux -
         meaning that the PID in the file will be an ascii string part of
         /proc/PID/status which is written in the file
         """
         shutil.copyfile(core, newcore)
-        with open(newcore, "r+") as f:
+        with open(newcore, "rb+") as f:
             f.seek(offset)
-            self.assertEqual(f.read(5), oldpid)
+            currentpid = f.read(5).decode('utf-8')
+            self.assertEqual(currentpid, oldpid)
 
             f.seek(offset)
             if len(newpid) < len(oldpid):
                 newpid += " " * (len(oldpid) - len(newpid))
             newpid += "\n"
-            f.write(newpid)
+            f.write(newpid.encode('utf-8'))
 
     def test_deeper_stack_in_minidump_with_same_pid_running(self):
         """Test that we read the information from the core correctly even if we
         have a running process with the same PID"""
         try:
             self.do_change_pid_in_minidump("linux-x86_64_not_crashed.dmp",
                                            "linux-x86_64_not_crashed-pid.dmp",
                                            self._linux_x86_64_not_crashed_pid_offset,