Replace the monitoring lock with a configurable semaphore

[joy4eg]

Currently, we have just a bool flag that controls the parallel execution.

The idea is to replace that flag with a semaphore and init it from the user's config.

Something like:

max-concurrent-monitorings: X

Where X may be one of:

• 0 - Disable that feature (disable-monitoring-lock:true as for now)
• 1 - Use one concurrent request at a time (disable-monitoring-lock:false as for now)
• N - The actual number to use.

As for the implementation perspective, it will require changing some default values/config parsing AND replacing current mutex with the actual semaphore.

[TwiN]

Why?

The reason why I configured it to allow one endpoint at a time from being evaluated concurrently is that more than that may provide inaccurate results (e.g. response time).

I'm not against considering it, but I'd like your reasoning

[joy4eg]

Scalability.

With 200+ endpoints, when the average response time is 80ms-200ms (due to different regions), having frequent checks (<=30sec) could be challenging when disable-monitoring-lock:false.

Semaphore will give us the ability to tune how many checks we allow rather than allow all (as for now), and thus may provide more accurate results.

[laurentpellegrino]

I have a similar use case and would appreciate such an option.

[TwiN]

Makes sense!

[TwiN]

While working on #1239, I decided this was going to be a necessity. It will be implemented alongside suites.