Remove any unbounded waits #122
Description
This a necessary step for decent real-time performance. I don't believe we have any issues here aside from allocation, which we're moving toward removing, and RwLock. We will need to move to a lock implementation that is completely fair. This can be done as part of #38. This issue is really just a top level tracker for any other cases of unbounded delay.