fix: Rare flush timeout when called in tight loop

Fix an edge case that flush could keep timing out when called in a tight
loop. The test case testFlush_WhenNoInternet_BlocksAndFinishes does this
and sometimes fails with 'Flush should not time out'. This is fixed now,
by removing the duplicate setting of isFlushing to NO in two different
synchronize blocks.

CHANGELOG.md

@@ -24,6 +24,7 @@ This bug caused unhandled/crash events to have the unhandled property and mach i
 - Don’t force cast to `NSComparisonPredicate` in TERNARY operator (#4232)
 - EXC_BAD_ACCESS in SentryMetricProfiler (#4242)
 - Missing '#include <sys/_types/_ucontext64.h>' (#4244)
+- Rare flush timeout when called in tight loop (#4257)
 
 ### Improvements
 

Sources/Sentry/SentryHttpTransport.m

@@ -209,10 +209,6 @@ - (SentryFlushResult)flush:(NSTimeInterval)timeout
 
     intptr_t result = dispatch_group_wait(self.dispatchGroup, dispatchTimeout);
 
-    @synchronized(self) {
-        self.isFlushing = NO;
-    }
-
     if (result == 0) {
         SENTRY_LOG_DEBUG(@"Finished flushing.");
         return kSentryFlushResultSuccess;