Disable parallel GC by default

Not using parallel GC seems to reduce avg pauses by ~0.1s on
EvaluateAndApplyChanges. Avoiding creating the thread pool should
help app launch slightly and reduce memory ussage.

Change-Id: Iebec2a17701c76e4145b41d7c0b4f6dd17806efa
diff --git a/runtime/gc/collector/mark_sweep.cc b/runtime/gc/collector/mark_sweep.cc
index bb8d876..e65d1f1 100644
--- a/runtime/gc/collector/mark_sweep.cc
+++ b/runtime/gc/collector/mark_sweep.cc
@@ -740,11 +740,7 @@
   if (heap_->GetThreadPool() == nullptr || !heap_->CareAboutPauseTimes()) {
     return 1;
   }
-  if (paused) {
-    return heap_->GetParallelGCThreadCount() + 1;
-  } else {
-    return heap_->GetConcGCThreadCount() + 1;
-  }
+  return (paused ? heap_->GetParallelGCThreadCount() : heap_->GetConcGCThreadCount()) + 1;
 }
 
 void MarkSweep::ScanGrayObjects(bool paused, uint8_t minimum_age) {
diff --git a/runtime/gc/heap.h b/runtime/gc/heap.h
index 2f62798..066b4c5 100644
--- a/runtime/gc/heap.h
+++ b/runtime/gc/heap.h
@@ -145,6 +145,8 @@
   static constexpr double kDefaultHeapGrowthMultiplier = 2.0;
   // Primitive arrays larger than this size are put in the large object space.
   static constexpr size_t kDefaultLargeObjectThreshold = 3 * kPageSize;
+  // Whether or not parallel GC is enabled. If not, then we never create the thread pool.
+  static constexpr bool kDefaultEnableParallelGC = false;
 
   // Whether or not we use the free list large object space. Only use it if USE_ART_LOW_4G_ALLOCATOR
   // since this means that we have to use the slow msync loop in MemMap::MapAnonymous.