refactor: dispatch system

[mehah]

Dispatch has been refactored to be centralized in just one thread. Previously, each task was sent to the threads that were free, which could cause bottlenecks because they were stuck with mutexes, causing queues depending on how many threads the cpu had.

small illustration of a possible problem:

Apart from what was mentioned above,, the scheduling class has been removed and we now have new methods in dispatch, addEvent_async, scheduleEvent and cycleEvent.

Note

• addEvent_async(func, context_id)
  • With this method it is now possible to create an asynchronous event that will be executed before all synchronous events.

Questions

• Why Scoped Object
  • there is a big cost in allocating(new) and deallocating(delete) an object from memory and since we don't need to manage the lifetime of the object or share it, it doesn't make sense to use smart pointer.

Benchmark

• Scoped Object vs Smart Pointer
  

struct Teste {
	int i;
};

Benchmark bm;
for (uint_fast32_t i = 0; i < 999; ++i) {
	std::vector<Teste> list;
	for (uint_fast32_t d = 0; d < 99999; ++d) {
		list.emplace_back(d);
	}
	list.clear();
}

g_logger().info("Scoped Object {}ms", bm.duration());

bm.reset();
bm.start();

for (uint_fast32_t i = 0; i < 999; ++i) {
	std::vector<std::unique_ptr<Teste>> list;
	for (uint_fast32_t d = 0; d < 99999; ++d) {
		list.emplace_back(std::make_unique<Teste>(d));
	}
	list.clear();
}

g_logger().info("Smart Pointer {}ms", bm.duration());
```</details>

[sonarqubecloud]

Kudos, SonarCloud Quality Gate passed!   

0 Bugs
0 Vulnerabilities
0 Security Hotspots
14 Code Smells

0.0% Coverage
0.3% Duplication