this means that, for example, if a gate has three outgoing edges with 33% percent likelihood each, one must be chosen at random (note that if edges have different likelihoods this will entail weighted sampling of the edges) and then it must push with no chance of failing.
note that if the triggering Gate first chooses an edge and then sends push!to it, there's still a chance that this push will fail because this edge has 67% chance of failing (the edge doesn't know that it is one of three gates and that one of them must push)
this means that, for example, if a gate has three outgoing edges with 33% percent likelihood each, one must be chosen at random (note that if edges have different likelihoods this will entail weighted sampling of the edges) and then it must push with no chance of failing.
note that if the triggering Gate first chooses an edge and then sends
push!to it, there's still a chance that this push will fail because this edge has 67% chance of failing (the edge doesn't know that it is one of three gates and that one of them must push)