[Autonomous] Stuck Resolution

[nukkeldev]

Building on #91, and local/global navigation, we now need to be able to resolve when the rover is stuck, however that may be. This issue documents possible architectures and integration methods I've thought of (and/or drafted) while simulation and the rover are in the works.

The first approach I'd consider would be to do as little work as possible and rely on the behavior of how costmap changes propagate to the local planning. Assuming the behavior of local planning backtracks if the current cell grows impassable, ideally that behavior would resolve most stuck reasons that are resolvable (i.e. not flipped on it's back).