[Clang][IFS][Test] Work around in-process cc1 ASAN issues #2.

Using the same strategy as c38e425.

D69825 revealed (introduced?) a problem when building with ASan, and
some memory leaks somewhere. More details are available in the original
patch.

Looks like we missed one failing tests, this patch adds the workaround
to this test as well.

(cherry picked from commit e174da4)

clang/test/Driver/cl-showfilenames.c

@@ -2,11 +2,19 @@
 // target Windows.
 // REQUIRES: x86-registered-target
 
-// RUN: %clang_cl --target=i686-pc-win32 /c /Fo%T/ /showFilenames -- %s 2>&1 | FileCheck -check-prefix=show %s
-// RUN: %clang_cl --target=i686-pc-win32 /c /Fo%T/ /showFilenames -- %s %S/Inputs/wildcard*.c 2>&1 | FileCheck -check-prefix=multiple %s
+// NOTE: -fno-integrated-cc1 has been added to work around an ASAN failure
+//       caused by in-process cc1 invocation. Clang InterfaceStubs is not the
+//       culprit, but Clang Interface Stubs' Driver pipeline setup uncovers an
+//       existing ASAN issue when invoking multiple normal cc1 jobs along with
+//       multiple Clang Interface Stubs cc1 jobs together.
+//       There is currently a discussion of this going on at:
+//         https://reviews.llvm.org/D69825
 
-// RUN: %clang_cl --target=i686-pc-win32 /c /Fo%T/ -- %s 2>&1 | FileCheck -check-prefix=noshow %s
-// RUN: %clang_cl --target=i686-pc-win32 /c /Fo%T/ /showFilenames /showFilenames- -- %s 2>&1 | FileCheck -check-prefix=noshow %s
+// RUN: %clang_cl -fno-integrated-cc1 --target=i686-pc-win32 /c /Fo%T/ /showFilenames -- %s 2>&1 | FileCheck -check-prefix=show %s
+// RUN: %clang_cl -fno-integrated-cc1 --target=i686-pc-win32 /c /Fo%T/ /showFilenames -- %s %S/Inputs/wildcard*.c 2>&1 | FileCheck -check-prefix=multiple %s
+
+// RUN: %clang_cl -fno-integrated-cc1 --target=i686-pc-win32 /c /Fo%T/ -- %s 2>&1 | FileCheck -check-prefix=noshow %s
+// RUN: %clang_cl -fno-integrated-cc1 --target=i686-pc-win32 /c /Fo%T/ /showFilenames /showFilenames- -- %s 2>&1 | FileCheck -check-prefix=noshow %s
 
 
 #pragma message "Hello"