Update no-inlining for a clang-cl world

With VC++ marking a function as non-optimized with MSVC_DISABLE_OPTIMIZE would
prevent it from being inlined, however with clang-cl that does not work,
leading to confusing crash stacks.

The fix is to use NOINLINE instead of MSVC_DISABLE_OPTIMIZE, which is
more explicit and specific anyway.

Bug: 873359
Change-Id: Ieb5c29e3212596f928a636d15143d717bd2c744c
Reviewed-on: https://chromium-review.googlesource.com/1187265Reviewed-by: Alexander Timin <altimin@chromium.org>
Commit-Queue: David Bienvenu <davidbienvenu@chromium.org>
Cr-Commit-Position: refs/heads/master@{#585831}

base/task/sequence_manager/sequence_manager_impl.cc

@@ -744,11 +744,9 @@ SequenceManagerImpl::GetMetricRecordingSettings() const {
   return metric_recording_settings_;
 }
 
-MSVC_DISABLE_OPTIMIZE()
-bool SequenceManagerImpl::Validate() {
+NOINLINE bool SequenceManagerImpl::Validate() {
   return memory_corruption_sentinel_ == kMemoryCorruptionSentinelValue;
 }
-MSVC_ENABLE_OPTIMIZE()
 
 void SequenceManagerImpl::EnableCrashKeys(
     const char* file_name_crash_key_name,