Fix race in event_tracer/repeat unit tests

[alan-george-lk]

Context: occasionally the CI runs would hang indefinitely on EventTracer.* based tests. This smelled like a race condition.

Root cause:

Race identified in src/trace/event_tracer.cpp. The writer thread waits on a CV under g_mutex:

std::unique_lock<std::mutex> lock(g_mutex);
g_cv.wait(lock, [] { return !g_event_queue.empty() || g_shutdown_requested.load(); });

But StopTracing() was modifying g_shutdown_requested without holding the g_mutex.

g_shutdown_requested.store(true);   // outside the lock
g_cv.notify_one();
g_writer_thread.join();             // <-- blocks forever

Even though that individual variable is atomic, per the CPP docs on condition variables:

Even if the shared variable is atomic, it must be modified while owning the mutex to correctly publish the modification to the waiting thread.

Solution was to simply do just that in StopTracing().

Testing

Verified locally and in CI with 1000 repeats of all unit tests (single run prior).

Also updated this stage to run all unit tests 100 times, but keep GTest output minimal to avoid spam.

[stephen-derosa]

nice.

[stephen-derosa]

LGTM -- thank you!