feat: implement trigger timeout on aperiodic tasks

[doudou]

A.k.a. "port driven with timeout"

[pierrewillenbrockdfki]

This basically allows to run setTimeout_us on a FileDescriptorActivity, explicitly only allows a 0 setting for PeriodicActivity, SequentialActivity and SlaveActivity(and does nothing else), and for the other Activities(and FileDescriptorActivity), when the associated thread is not active, running or periodic, allows for aperiodic wakeups calling the Activities loop() or step() implementation.

This all makes sense, even though the code base is confusingly using a variable named task for what is a Thread or, at best an Activity, not related to the TaskCore and its similar named methods.

The sem_timedwait being tied to CLOCK_REALTIME is just sad, but it is better than nothing.

Thanks to this, i learned that i was rather wrong about PeriodicActivity accepting ::trigger from watched ports...

[doudou]

This basically allows to run setTimeout_us on a FileDescriptorActivity, explicitly only allows a 0 setting for PeriodicActivity, SequentialActivity and SlaveActivity(and does nothing else), and for the other Activities(and FileDescriptorActivity), when the associated thread is not active, running or periodic, allows for aperiodic wakeups calling the Activities loop() or step() implementation.

On FileDescriptorActivity, it makes a common API for an API that already existed.

For Activity when it is not periodic (i.e. port-driven), it allows to get a time-based wakeup.

For the others, it keeps their properties, which is unthreaded for SequentialActivity and SlaveActivity (which means we can't wake them up), and purely periodic for PeriodicActivity and Activity with a period.