Remove documentation for multiple invocations

This is no longer applicable as --output_dir defaults to a temporary
directory instead of a shared default.

Also fixing a unittest being broken on Windows after the last commit
(hardcoded ./ instead of os.path.join).

This should be the last remaining step to fix #54 and #57.
diff --git a/README.md b/README.md
index b07a194..2d51c8a 100644
--- a/README.md
+++ b/README.md
@@ -88,12 +88,3 @@
 `--serialize_test_cases` that runs tests within the same test case sequentially.
 While generally not providing as much speedup as fully parallel test execution,
 this permits such binaries to partially benefit from parallel execution.
-
-## Multiple Invocations
-
-When running multiple `gtest-parallel` instances, separate `--output_dir`
-locations should be used. `gtest-parallel` cleans out this directory on startup,
-to avoid mixing old stale logs with new ones. This will discard any logs from
-ongoing invocations, or otherwise fail (throw) because the files are in use
-(depending on platform). This is not enforced by a lock file as cleaning these
-up would require unnecessary user intervention.
diff --git a/gtest_parallel_tests.py b/gtest_parallel_tests.py
index 5693cd8..9adab79 100755
--- a/gtest_parallel_tests.py
+++ b/gtest_parallel_tests.py
@@ -315,7 +315,7 @@
       return 'C:\\' if sys.platform == 'win32' else '/'
 
     self.assertEqual(
-      './bin-Test_case-100.log',
+      os.path.join('.', 'bin-Test_case-100.log'),
       gtest_parallel.Task._logname('.', 'bin', 'Test.case', 100))
 
     self.assertEqual(