Review & update controller_interface documentation on realtime-safe methods

[bmagyar]

Methods that should be realtime-safe should have this documented in their docstrings

[dignakov]

What would you like to see, would something like //this method should be realtime-safe in the header be good enough?

[bmagyar]

Yeah!

[matthew-reynolds]

I think we should also add a "This method is non-realtime-safe" where appropriate to be explicit. Otherwise, no comment would be ambiguous between "Not yet documented" and "Non-RT-safe"

I think the ros(1)_control codebase does it that way, though I believe the coverage isn't 100%