[SofaSimulationCore] FIX Task scheduler memory leak

[alxbilger]

Fix memory leak in task scheduler:

• Add comments
• Add unit tests
• Use shared pointer instead of raw pointer:

The static raw pointer had two bad effects:

• Memory leak
• Task scheduler not being stopped

Note: The current API for the task scheduler is very confusing. I suggest to clarify the expected behavior.

---

By submitting this pull request, I acknowledge that
I have read, understand, and agree SOFA Developer Certificate of Origin (DCO).

---

Reviewers will merge this pull-request only if

• it builds with SUCCESS for all platforms on the CI.
• it does not generate new warnings.
• it does not generate new unit test failures.
• it does not generate new scene test failures.
• it does not break API compatibility.
• it is more than 1 week old (or has fast-merge label).

[fredroy]

[ci-build][with-all-tests]

[fspadoni]

Good to use a smart pointer and fix the memory leak.

What about using a unique_ptr ?
The task scheduler should manage itself and no other class should control its lifetime storing a refernce to it.

[alxbilger]

@fspadoni you're right, I'll change it

[alxbilger]

@fspadoni Is the task scheduler designed to have local schedulers? For example, can I have a task scheduler specific to a component? Is it restricted to have only one instance per derived type?

[fspadoni]

@fspadoni Is the task scheduler designed to have local schedulers? For example, can I have a task scheduler specific to a component? Is it restricted to have only one instance per derived type?

There is only one global instance of the task scheduler class and this is shared between all components.
Any components can create and submit tasks to task scheduler.
The tasks are queued and process by the worker threads.
The task scheduler manages the worker threads (only one per cpu core by default)

[alxbilger]

@hugtalbot @guparan I am not sure I just need to switch from EXPECT_EQ to EXPECT_NE. When I'll refactor the interface of the TaskScheduler, a large majority of the test will not make sense. I suggest to merge this PR now, and I'll change the tests in my next PR.

[alxbilger]

[ci-build][with-all-tests]

[alxbilger]

I removed completely the unit tests. I'll add them later in another pull request.

[hugtalbot]

Looks fine to me but I am really no MT / TaskScheduler expert