Adjusting libcosimc to fiberless libcosim

[ljamt]

Adapting to the libcosim api change.

[markaren]

We can probably remove fibers in libcosimc altogether?

[ljamt]

We can probably remove fibers in libcosimc altogether?

Yes, but then we'll need to rewrite cosim_execution_start to spin off a thread for instance..

[markaren]

auto task = std::thread ([execution]() {
                execution->simulate_result.set_value(execution->cpp_execution->simulate_until(std::nullopt));
            });
            execution->t = std::move(task);

where simulate_result is an std::promise might do the trick. I wanted to test, but the dlls are not placed correctly so I can't run the code and I don't want to debug this. execution_async_health_check needs to go though. Otherwise just keep this code for now.

[restenb]

If the fiber is only used to spawn the main thread, which now runs most of it's work on separate hardware threads, I'm not sure I see the immediate need to remove fibers here as well, at least if it would demand further changes in libcosim.

The rationale for removing fibers from libcosim was basically that they weren't being used, and just added extra weight, wasn't it? The usage here seems OK to me.

[markaren]

If std::promise can do the same job, we don't need the extra dependency.

[ljamt]

If std::promise can do the same job, we don't need the extra dependency.

I agree, but suggest to solve that as a separate issue.