Originally from #826, but deferred from minimum-viable-product.

The current implementations do nothing, which we've found to be good enough for Tor simulations. They are probably usually good enough in general; generally these are backstops to prevent deadlock in the case that a thread or process exits without releasing its locks.

Otoh the deadlocks that would occur in a situation where these are needed may be difficult to debug, especially if encountered by someone who doesn't know or remember that these aren't implemented properly. In some cases like this a warning for debugging purposes would suffice, but in this case it'd be too noisy since we'd see the warnings in every Tor simulation.