[VM] - put the trace message which states a safepoint is taking too long

under a flag so that it is printed only when needed.
       Addresses issue 34933.

Bug: 34933
Change-Id: Idf77dd9492498d709384b51af66c464ae8c3fe00
Reviewed-on: https://dart-review.googlesource.com/c/81521
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Commit-Queue: Siva Annamalai <asiva@google.com>

runtime/vm/heap/safepoint.cc

@@ -9,6 +9,8 @@
 
 namespace dart {
 
+DEFINE_FLAG(bool, trace_safepoint, false, "Trace Safepoint logic.");
+
 SafepointOperationScope::SafepointOperationScope(Thread* T) : StackResource(T) {
   ASSERT(T != NULL);
   Isolate* I = T->isolate();
@@ -109,7 +111,7 @@ void SafepointHandler::SafepointThreads(Thread* T) {
       Monitor::WaitResult retval = sl.Wait(1000);
       if (retval == Monitor::kTimedOut) {
         num_attempts += 1;
-        if (num_attempts > 10) {
+        if (FLAG_trace_safepoint && num_attempts > 10) {
           // We have been waiting too long, start logging this as we might
           // have an issue where a thread is not checking in for a safepoint.
           OS::PrintErr("Attempt:%" Pd " waiting for %d threads to check in\n",